Impact of innovation: Context

Here we explore the technology areas that our working group felt have already had, or are starting to have, an impact in paediatric care.

In September 2019, our working group met up and agreed a number of technology areas that have already had, or are starting to have, an impact in paediatric care. For the following case studies we looked at the background research, how the innovation is being used, and current limitations and obstacles to use. We also thought about recommendations and solutions, which are discussed later in this report.

Not covered in detail here, but also identified as important by our group, are additive manufacturing/ 3D printing, assistive technology and diagnostic hardware and software. We also can’t forget the more “obvious” technology that has been transformational in the way we deliver care and live our lives – digital and communication. Lots of the studies in our literature review [1]  were digital or telecommunication interventions, so we suggest looking at that for more information about their impact.

Below we have summarised the horizon scanning done by our members and experts. We have also included a national case study of paediatric innovation from the UK.

Mobile health/apps provide a resource for parents to support their children with different diseases and health issues. >90% of young adults have a smartphone – apps could therefore be one of the best ways to share information with this population group. They can improve health outcomes, reduce costs and empower patients. However, they vary in quality, and there are a limited number of apps for use specifically in paediatrics.

Current research

Services currently provided by apps include disease information, healthcare education, pharmaceutical information, online patient check-up, doctors’ information, health monitoring and patient health records.[2] Apps are very flexible and can be used over a variety of settings/to provide different services. One example is paediatric palliative care, where apps can be developed for calming, relaxation and mindfulness.[3] Another is for parents with infants in neonatal intensive care units (NICU) and paediatric intensive care units (PICU) such as the VCreate application.[4] These apps can be helpful in communicating information to parents/carers and ensure communication needs are met.

Currently NHS have an apps library[5] and support the use of apps – 14 apps in total on NHS apps library for child health. ORCHA (Organisation for the Review of Care and Health Apps) reviews apps for many NHS bodies in the UK, at a national and local level and a wide range of international health and care systems.[6] ORCHA independently provides information and scores apps using a paediatric team with NHS paediatricians evaluating apps and the research behind them. A review process checks that information is aligned with medical bodies, national organisations associated with the app, the medical lead for the apps, and quality of research or beneficial evidence.

Limitations and obstacles to use

There is a lack of scientific information measuring the diagnostic value of health apps in current medical literature. Apps could be used to monitor healthcare, but currently most paediatric apps are limited to diagnosis of diseases.[7] In addition, few studies have explored the quality and reliability of apps specifically for children and young people.

There is a high cost to set up apps, and keep them running, including continued evaluation and updating content and functionality. There also remains a variety of quality and credibility of apps, despite NICE evidence standards for medical and health apps.[8] Parents and healthcare professionals need to be aware of the issues with quality and credibility.

There are ethical issues in relation to development and updating apps – these include data sharing, and ensuring children and young people don’t have the opportunity to accidentally share medical information with other sites or social media, as it often the case with adult health apps. Designers of apps need to be aware of the adverse effects/consequences of sharing incorrect information. In the future, we need concise guidelines for developers, especially considering changes in treatment frequently occur and app systems need to constantly be updated. There are also ethical considerations with storage and sharing of information.[9], [10]

Apps need to be available on both iOS and Google play app stores and we must remember you can’t access an app if you don’t have a device to do so in the first place. Web based apps are emerging which can be accessed through web browsers and therefore allow integration with other computer systems and can use the most modern web capabilities. Web apps are likely to provide equality around a wider reach, customisation and digital access to apps.

Key questions for future research

  • What is the current user opinion for use of apps in paediatric healthcare?
  • In the future, will apps be necessary/be used?
  • What apps need to be developed to support paediatric healthcare?

Artificial Intelligence (AI) has been described as a set of advanced technologies that enable machines to do highly complex tasks effectively, which would require intelligence if a person were to perform them.[11], [12] Inevitably, the definitions of “advanced” and “intelligence” are broad and undefined.

Perhaps the most well-known current application of AI is with the advent of predictive text in smartphones and self-driving cars.

The Topol Review, commissioned in 2019 to describe the future of the digital NHS, highlighted artificial intelligence as having the potential to transform the delivery of healthcare through several methods:[13]

  1. Streamlining workflow processes
  2. Improving the accuracy of diagnosis
  3. Personalising treatment
  4. Helping staff work more efficiently and effectively

AI is not expected to develop to the extent that it will fully replace clinicians;[14] however its scope is likely to expand rapidly to the point of automating repetitive or analytical tasks beyond human capabilities. This then enables the healthcare workforce to focus wholly on tasks considered “uniquely human” such as interaction and care.[15] [16]

Current research

AI has already established itself as being effective at image classification and interpretation, with several real-world clinical uses. Specialties such as radiology, pathology, dermatology and ophthalmology have benefited the most from the ability to classify radiographic images or pathology slides. [17] Teams at Google Health and its DeepMind subsidiary have published significant data around this:

  1. An AI model was used to review a patient’s chest CT scans to predict the risk of lung cancer. The algorithm was trained using 6,716 scans and validated using 1,139 cases, compared to a gold standard of the interpretations of six radiologists.[18] The AI model was found to be non-inferior where no previous imaging was available and had achieved 11% less false positives and 5% less false negatives than the radiologists when a prior scan was available for comparison.
  2. More recently, a similar model was trained on de-identified mammograms from 76,000 women in the UK and 15,000 women in the US to attempt to learn the signs of breast cancer in the scans.[19] It was then evaluated against a separate set of 28,000 women across both countries and compared to expert radiologists. The model achieved a significant reduction in false positives and false negatives (up to 5.7% and 9.4% respectively in the US) when compared to the radiologists.
  3. AI models were used on various retinal image collections, including the UK Biobank collection of 114,205 fundus images and in collaboration with Moorfields Eye Hospital in London, with the aim of detecting retinal disease.[20] [21] The model has been trained to detect anaemia and signs of age-related macular degeneration with remarkable accuracy.

AI also offers the potential to automate straightforward transactional conversations with humans (patients) to manage scheduling or simple structured questionnaires without requiring the user to engage with a tablet, smartphone or PC.[22]

Finally AI offers some potential in the areas of clinical decision support. The team at ADA Health have conducted research on “Probabilistic Diagnostic Decision Support systems” to aid the process of rare disease diagnosis, with the aim of improving the time taken to correctly diagnose rare conditions.[23]

However, as these case studies show, few are specific to children and young people. This doesn’t mean that AI won’t be beneficial for children but we do need to create the conditions whereby AI research in children can flourish – as there may be unintended consequences if we don’t thoroughly investigate all possible outcomes. The below studies show there is great promise in this technology.

Case studies of use

In 2010 Zargis Medical brought to market and automated auscultation device that could determine which murmurs in children were pathological and thus managed referral for Echocardiogram better than a paediatrician.[24]  Zargis medical has since gone out of business but similar technology is in development currently[25] and is being focused on low to middle resource settings to reduce unnecessary referrals.

Aberdeen university are involved in developing a system called BabyTalk that can automatously create written summaries for babies on the NICU.[26] [27] They trialled their BT-45 system which summarises activity over the preceding 45 minutes and found it supported effective decision making. Currently they report that their artificially generated notes are not yet as good as those of clinicians and lacking narrative structure but work is ongoing to improve this. They also plan to develop tools that can create summaries for parents to aid communication.

Some AI is already in routine paediatric practice. At the time of writing, two AI-enabled hybrid-closed-loop insulin pumps are in routine use in NHS paediatric diabetes clinics, where AI-algorithms use data from continuous glucose monitors to change insulin delivery in real-time. Some patients and families go past this and use ‘DIY’ systems to fully automate their pumps using community-written algorithms, currently outside NHS governance structures.

Limitations and obstacles to use

Despite many people’s perceptions AI is not a panacea: its limitations must be understood. For example, when it presents as a deployment of machine learning the outputs of the system will be based on the dataset that was used to train the system, it does not have intuition! Thus, if it was trained using data drawn from a North American population it may not work as well on an English population. This is a limitation that must be understood, but it is also a strength: given an appropriate data set, it can be trained to deliver for the English population and any other population. The national location would just become another parameter in the dataset. It is an ideal tool for the delivery of personalised medicine.

Many concerns are being raised about current deployments of AI systems. The most obvious of these is the GP at Hand service for primary care deployed by Babylon.[28] The issues that arise from changes in pathways/business models, change management and the technology itself are frequently conflated. This is unhelpful and commentaries should always be reviewed with a view to understanding which these areas are relevant to the issues raised.

There are genuine concerns about accountability and governance when AI is deployed in the healthcare system. There needs to be a significant focus on understanding the issues at play and the development of frameworks through which the risks presented by AI technology can be evaluated against the benefits.

The term “Big Data” has been described as a collection of data or information that is too large or complex to analyse using traditional methods of database software, and is instead characterised by the “three Vs”: volume, velocity and variety.[29], [30], [31]

Volume refers to the amount of data generated in real time, for example through computer software use, sensor devices or web streams or data feeds. Velocity refers to the fast rate at which the data is received, and variety refers to the many types of data that are made available largely in unstructured forms.

Big Data already has a huge number of use cases in a variety of industries.[32]. [33] Big Data has tremendous potential for research and clinical use. Currently a vast amount of data is generated through digitization of patient records, medical imaging, laboratory results, medical devices and sensors, and even newsfeeds, “App” data and social media streams available on the internet.[34]

Current research

Big Data is already being used to address numerous questions and issues in healthcare:[35], [36], [37]

Research into clinical and operational efficacy

Big Data can be used to determine more clinically relevant and cost-effective ways to treat and diagnose patients. Certain outcomes may be predicted or estimated based on historical data, such as estimated number of A&E attendances, predicted hospital length of stay, or determining which patients are at risk of operative complications.

Public health applications

Disease patterns can be analysed to track outbreaks and transmission with the aim of improving public health surveillance and response. Data can be used to identify needs, provide services and predict and prevent population level crises.

Population level risk stratification

Patient profiles based on large volumes of population-level data can be used to proactively identify patients who might be at risk of developing certain diseases or conditions, allowing for better targeting of preventative care or targeted screening and intervention. By linking data between healthcare providers new incites can be gained into the care of patients with rare conditions which previously have been difficult to study in an individual centre.[38]

Patient profile level risk stratification

Likewise, patient-level deep data including but not limited to wearable device data, genomics, proteomics and transcriptomics can reveal useful information about a person’s risk level and help predict future disease, thus enabling early preventative care, targeted screening or intervention. The introduction and increasing use of electronic health records (EHR) has provided new opportunities to use the large volumes of data being generated by providing routine care to develop early warning tools to help clinicians identify patients at risk of deterioration.[39], [40]

Case studies of use

Alder Hey Children’s Hospital collaborated with IBM to personalise and improve patient experiences when attending hospital.[41], [42] By collecting data and making it accessible, hospital staff were able to identify clinic trends, predict attendances and personalise the patient experience making their young patients feel more comfortable with visiting this hospital. Their scope has since expanded, with future targets including identifying trends that affect patient flow and mitigating these, matching patients with suitable clinical studies and using interactive applications to help manage chronic illnesses and prevent hospital admissions.

Bradford Teaching Hospitals have partnered with GE Healthcare to create a centralised Command Centre to improve how they plan and manage patient flow across the hospital.[43], [44] Such facilities were already established at the Johns Hopkins Hospital in Baltimore, USA, resulting in a 30% improvement in time to bed allocation and a 70% reduction in delays from post-procedural delays in transfer.[45] They are able to use vast amounts of real-time data from hundreds of systems to anticipate and resolve bottlenecks, displayed on a number of screens likened to an air-traffic control centre. They intend to use this system to predict likelihoods of attendances and length of stay, to then increase the proportion of patients treated in A&E within four hours and ensuring patients are directed to the most appropriate wards.[46] This then helps predict and prevent system bottlenecks in patient flow, thus improving patient safety and driving down costs.

The National Neonatal Audit Program is an example most of our members will be familiar with. The national clinical audit run by the Royal College of Paediatrics and Child Health that was established in 2006. It collects anonymized data from the patient record system in all participating neonatal units and collates it into a national database. It provides an opportunity for individual units to compare locally practice against national norms and gives a large dataset for new research.[47], [48], [49] It is a model of what is possible when data is effectively shared between organisations.

There is ongoing work to use data from electronic health records for the early identification of paediatric AKI. Numerous models have been designed for use in adult practice taking advantage of the confluence of both increasing patient data and a universally agreed definition of AKI. These can provide real time, accurate alerts to clinicians as soon as a patient meets the diagnostic criteria. Evaluation of the various systems shows these alerts work best when providing relevant recommendations to the treating clinicians.[50], [51] Similar work is ongoing regarding the identification of Sepsis. The Nervecentre platform, at Nottingham University Hospitals and University Hospitals of Leicester NHS Trust integrates both pathology results and vital signs to help early identification of sepsis.[52]

Limitations and obstacles to use

Privacy and security

Legal and regulatory aspects of privacy protection and data sharing.[53] A current concern related to the advent of Big Data is data privacy, namely the concern that organisations handling the data might use it for profit, such as in the sale of data to third parties, or for discrimination such as in the case of health and life insurance and employment selection.

Data quality and compatibility[54]

Data generated from healthcare is often heterogeneous, distributed across many institutions, locations and held within a variety of proprietary software. This data is affected by the “Garbage In-Garbage Out” principle of data quality, and poor clinical coding often results in limited utility of datasets. Significant use of Big Data will require some degree of standardisation and interoperability in data.

Training and educational requirements of big data[55]

Effective data management requires specialist training in data science and health informatics. Use of this data requires extensive knowledge in how it can be processed and utilised, often using machine learning algorithms. Even clinicians require training in health informatics, as collectors and curators of data (18). Currently few training programs and job roles exist (although this is increasing), and health informatics is not included in most medical school curricula.

Existing IT infrastructure

The quality of current IT through the various health care environments is variable with staff reporting difficult or clunky systems and many areas still being significantly paper based.[56], [57]

Key questions for future research

  • How can we use big data to improve the care that paediatric patients receive?
  • What are the main barriers preventing these innovations from being developed?

Globally 350 million individuals are affected by serious genetic diseases (3.5m in the UK). One third of those affected will not live to see their fifth birthday, meaning the burden of genetic disease is greater for young children and their families. Many more children and young people (0-25-years of age) will develop serious physical and mental health conditions, many of which will be genetically determined.

Current research

There is strong evidence that whole genome diagnostics can improve care for seriously ill children, including children and young people with cancer. Beginning at birth, and focusing on seriously ill children in intensive care, we can use whole genome screening (WGS) for rapid diagnosis and to develop a prospective, living database that is constantly evolving and shaping our understanding of how genes affect lifelong healthy ageing. The UK is uniquely placed to deliver this type of research owing to the robustness of data available for individual children across health and education. This has led to recommendations for implementation of whole genome sequencing (WGS) in the NHS 10-year plan.[58]

Current research focuses around four areas, which are interlinked:

  1. Improving diagnosis
    Rapid diagnosis of serious genetic disease
  2. Precision medicine
    Targeted therapy applied to genetically-modified cohorts
  3. Health economy
    Improvements in care and reduce costs
  4. Family centred
    Engage families to measure the benefit and impact of diagnosis

It is anticipated that future paediatricians will use the genome and other biomarkers for early detection of disease, as well as prediction and prevention of illness by identifying an individual’s health risks and personalised practices to promote a healthy childhood and improved health outcomes throughout the lifespan. An integrated clinical and research approach is needed to show how this can inform practise across diverse populations.

Research demonstrates the use of WGS for early detection of serious genetic diseases in NICU and PICU.

The next ten years will see great progress in decoding the human genome leading to the “Neonatal Screen of the Future.”, providing rapid genetic diagnosis for conditions that present as life-threatening illnesses in NICU and PICU (~100,000 per annum in UK), potentially avoiding the need for invasive and expensive investigations. This approach empowers clinicians and patients, as early diagnoses will allow for earlier access to potential treatments.

With progress in the field, reduced costs of WGS and growing expertise in the NHS and UK HEIs, by 2030 genomic medicine will expand to cover diabetes, cardiovascular, mental health and other common conditions that have their origins in children and young people (0-25-years). We will better understand the nature of resilience and how this can be augmented to delay disease onset or prevent disease from ever occurring.

By 2040, we expect that paediatricians will routinely use WGS information to structure healthcare of each child, and that benefits to children and families will be better established. As such, educational programmes in paediatrics and general practice will need to provide greater knowledge of genomic working, and promote training of medics and junior doctors to take on the important challenge for paediatrics.

Limitations and obstacles to use

Responsible management and storage of data is an important consideration when it comes to WGS. We can consider the advantages of housing complex data in the newly established NIHR Paediatric BioResource, for children and young people 0-25 years with physical and mental health conditions, to facilitate clinical benefit in the NHS, research opportunities for data science, precision medicine, industry interactions and health economic analysis. Like the UK BioBiobank, this resource can facilitate data linkage, access for research, recontact patients for re-consent into future research. The UK NIHR BioResource is intended to be a secure data platform to support (1) improved diagnosis, (2) data science and linkage, (3) personalised medicine, (4) benefits patients and families, ethically and psychodynamically, and (5) the health economy.

A sensor is a device that “detects or measures a physical property and records, indicates, or otherwise responds to it”.[59] Sensors are ubiquitous in healthcare and are widely used for example to measure blood pressure, heart rate, and blood oxygen saturation levels. Sensors are also used widely in medical laboratories to interrogate specimens and convert information into data that can be appreciated by clinicians.

Current research

Novel implementations of pre-existing technology

The first route to further develop the use of medical sensors is to take pre-existing sensor technology and apply this in novel ways.


This could involve applying technology such as heart rate sensors into smart watches in order to continuously monitor heart rhythms, or socks for babies that integrate blood oxygen saturation monitoring. This is pre-existing technology in that the sensors already exist and are widely used, but this is a new application of this device.

Relocated laboratory investigations

As our ability to shrink microprocessors and sensors into ever-smaller devices continues, it will be possible to perform investigations in places that heretofore have usually been performed in medical laboratories.

For example, GP practices could perform point-of-care testing on patients for haemoglobin, C-reactive protein and similar such biomarkers to help triage or aid in decision-making. There are already NICE MedTech innovation briefings[60] about such devices.

In the future, these devices could move even closer to the patients. For example, it might be an attractive proposition for parents of children with certain metabolic diseases to be able to perform finger-prick blood tests on their children and be able to call up an outreach clinician from the local metabolic service with the child’s blood glucose and blood ammonia results already available to help guide in decision making.

Novel sensors for pre-existing biomarkers

A recognised drawback in modern paediatric practice is that the acquisition of frequent blood samples in this patient population is the source of significant distress (to parents and children) and resource allocation (play specialists, junior doctors, nurses etc). If sensors could be designed to accurately run more tests on smaller blood volumes, and could be validated for capillary versus venous samples, it would obviate the need for phlebotomy on many patients.

Novel targets for medical sensors

It is comparatively more speculative to consider the development of new sensors because this depends on the development of new technology with underpinning biomedical research. Given the significant time lag for basic scientific research translating into products for sale, it may be that the research that will underpin products in 2040 is only in its early stages. It may be more helpful to consider a later stage of development – for example products which have been brought to market and have applied to NICE for permission to be used in the NHS.

Case studies of use

These examples, and others, are available on the NICE website.[61]

MedTech innovation briefing [MIB132]

This is a home point-of-care test for faecal calprotectin that can be used to assess inflammatory bowel disease. The patient uses the test and the information is transmitted to their clinician via smartphone.

MedTech innovation briefing [MIB110]

This is a device which measures blood glucose concentration by measuring the glucose in interstitial fluid via a flexible fibre which is pushed 0.5cm into the skin. People have described the application of this device as painless. It can be integrated with a dedicated reader or by a smartphone with near-field communication.

MedTech innovation briefing [MIB114]

This device tests for CRP and MxA (myxovirus resistance protein A) in primary care settings. It gives the GP extra information to judge whether the patient has a viral or bacterial infection.

Limitations and obstacles to use

Some obstacles to the widespread use of new sensors might include:

  1. Cost – an NHS increasingly short of money would not be able to adopt new technology that was expensive in its own right, unless it could be demonstrated that significant savings could be made through the use of this technology
  2. Brexit – the necessary legislation to bring these products to market in the UK may not exist, or it may be prohibitively expensive for companies to demonstrate that their products meet both European, and British regulations if we cannot align our medical device regulations to the European market.
  3. Lack of funding for primary research – if the UK research environment is not thriving and producing new targets for medical sensors then there can be no innovation in this area.

Key questions for future research

Our key question for the future is “what new routes for development of medical sensors could be explored to make medicine better for families and clinicians by 2040?”. This question can be divided into two main areas of research:

  1. Can we develop new sensors for existing biomarkers that reimagine the way that healthcare is delivered to children and young people? For example, home testing or GP point-of-care testing that will allow clinicians to make faster decisions and result in shorter admissions.
  2. Can we develop new biomarkers that will change the established management pathways for common or rare diseases in children and young people?

Virtual reality (VR) and augmented reality (AR) are two simulation models that are likely to transform medical education. VR provides a 3D and dynamic view of anatomical structures and allows the operator to interact with them. Recent advances in display systems, haptics and motion detection empower the user to have a realistic, interactive experience, enabling VR to be ideal for training in hands on procedures. This makes VR have limitless potential for teaching practical or interactive procedures and is likely to significantly influence medical training.

AR enables us to project virtual information and structures over physical objects. In practice this allows the trainer to enhance or alter the real environment. AR has the potential to improve training in the understanding of anatomical structures and physiology.

Current research

There is already emerging evidence validating the effects VR and AR have on medical education. Traditional High-fidelity VR based labs already exist in some hospitals in the UK, such as Nottingham University Hospital whilst at University Hospital North Midlands a unique Extended Reality Lab (ERL) has been developed.[62]

The main objective of VR is user’s immersion in virtual environment. For the experience to be close to the “real thing” structures in the virtual environment need to have high fidelity and the interaction of the user with the virtual environment created needs to be realistic. Such technology is available and is dependent on a number of technological parts to be successful. These include haptic devices, high-resolution audio-visuals, motion detectors, and of course fast and advanced processors that can receive, process and transmit with no lag.

Although there is significant overlap between AR and VR in respect to some technical aspects, AR differs from VR, as its target is not to construct a fully artificial environment but rather to superimpose computer-generated images on images of the real world.[63] AR therefore uses hardware that allows physical view of the surrounding environment to be visualised but these surroundings are enhanced with virtual images. Tablets, mobile phones, AR glasses, are most commonly used for running AR applications.

At present, AR is used widely in the clinical setting, especially during interventional procedures (CT/MRI guidance, visualization of paths), but perhaps its most promising application is in medical training and education. Anatomy, teaching with the 3D visualization of complex structures (such as complex congenital heart disorders) as well as physiology education by representing mechanisms in 4D (in space/ time dimensions) generate a lot of medical literature.[64]

A key challenge for VR and AR simulator is that it must adhere to certain quality standards in order to be suitable for medical education.

Numerous published papers assess the use of VR and AR simulators in clinical educational settings. The objective is to ascertain if the use of these modalities in medical education is valid and the skills learned transferable to a real clinical setting. Other educational aspects looked at how these technologies affect the student’s skills retention and learning curve. Any application of VR and AR technologies into the RCPCH curriculum is likely to need to take such evidence into account in order to help integrate evidence into practice in a way that it enhances current learning systems but also in a cost-effective manner.

High-resolution images, sound quality, haptic input/output, high-powered processors are vital in creating realistic VR/AR environments. High fidelity of structures such as vessel, vital organs, skin, bone and other tissues is integral to the creation of a successful virtual environment. These need to have properties of shape and size change in proportion to the degree of manipulation by the user, i.e. the capacity to change in relation to how it is manipulated. Without this the learned skills are likely to be inappropriate and not applicable in real life scenarios.[65]

To maintain quality in the AR/VR educational hardware ISO criteria have been established. These include an initial assessment of the educational needs and the formation of a VR/AR based solution. This obviously requires in-depth knowledge of clinical process that we need to simulate. The setting then needs to be validated, in order to examine if it meets the educational needs. For all this to create a high standard educational product technical experts, medical experts, human factor experts, and designers need to collaborate and form a diverse team that is able to design, evaluate, and upgrade the VR/AR products.[66]

Upon their release, clinically orientated VR and AR systems need to undergo validation in order to ascertain if they provide the educational effect they have been designed to provide. A number of ways of validating them have already been described in the literature.[67]

Once the educational validity of a VR/AR tool is ascertained, the tool needs to be assessed further against how it complements existing learning processes in a way that accelerates the acquisition of new clinical skills (thereby, arguably having the potential of shortening the training process). Stereotypical learning curve graphs depict the level of learned skill and the amount of time spent learning it (axis usually in time or number of times the specific procedure has been performed) Generally speaking educators aim for a curve with a steep curve with a high plateau (i.e. high level of skill within a short amount of time). The trend amongst most learning curves generated by those that acquired skills using VR/AR simulators have a low plateau but a steep shape.[68] This implies that the skill level learned is lower than with other modalities, but these skills can be learned fast, making the VR/AR platform ideal for the education of novices (but less so for the experienced clinician).

Limitations and obstacles to use

VR simulators are expensive to purchase, however they can sustain repeated use with low cost and require less personnel than other methods (animal models, actors or real patients). Furthermore AR/VR learning processes minimise the initial risk to the patient.

These technologies, however, are not without their drawbacks. If either the software or hardware malfunctions the virtual learning experience is ruined. Both VR and AR come with certain high-specification hardware requirements in order to retain satisfactory standards of simulation. 3D images, audio files need to emulate the real world realistically and have the ability to simulate real and abstract structures. Motion sensors must to be precise, sensitive and with no lag between user movement and screen projection. This is imperative so to enable the user’s that the visual field (size, shape, object angle) and auditory stimuli (volume, sound balance) to change in a realistic fashion. The student must not only be able to affect the virtual environment but also to be able to receive feedback via haptic stimuli. Haptic devices, such as joysticks, sensory gloves, and other specialized devices are currently in use but these can be costly.

Furthermore, high-performance computers are the cornerstone of VR/AR, as they are needed to process vast amounts of information with minimal latency. This needs to be coupled by fast Internet connection. All the above features are necessary in order to have a seamless immersion of the student in a virtual learning environment. These all come at a cost and are still relatively niche technologies in the world of medical education. These are likely to become more affordable as technology advances and VR/AR becomes more popular as a valid educational modality.[69], [70]

Wearables are sensor containing technology that can be worn on the body to monitor and collect data on physiological responses. Wearables can allow patients to self track, monitor, potentially diagnose or manage a medical condition or health need.

Wearable technology can be worn or attached to almost all areas of the body from headbands, eyewear, electronic skin patches, smart garments, baby-grows with sensors, activity trackers, e-footwear, smart watches, smart rings, smart jewellery and smart sports equipment such as smart cycle helmets. Wearables can collect data, assimilate data, interpret data (see the section on AI) and potentially assist the user by educating and influencing their behaviour, alerting them (or their carer or doctor) or treating them (for example diabetes wearables detecting glucose levels and directly instructing an insulin pump). Wearables can be connected to apps, phones, computers, medical records, and other wearables.

There are well over 500 million different wearable technologies,[71] the trajectory suggests this will double in the 2020’s, growing vastly by 2040 particularly as Bluetooth capabilities increase.

Young patients are more engaged with digital platforms as NHS England recorded in its Diabetes Prevention Programme.[72] Many girls use menstrual cycle apps but wearables can also track related menstrual cycle changes in the body: this is well developed for the adult population to target the fertility market, but could be used for peri-pubescent girls to help target onset of menstruation for those with intellectual disabilities or long term conditions where support is needed. Also wrist wear, particularly smart-watches and health and fitness trackers are popular amongst young people. Given many children have an elderly relative elderly with an emergency red button that contacts a care service if they fall, integrating trusted technologies that the public and NHS staff are already aware of give wearables a slightly higher trust potential than other technologies aiming to reach the market.

Current and future research

Research suggests that technology can accurately and reliably monitor the patient’s physiological status across a range of vital signs such as pulse, heart rhythm, respiratory rate, oxygen saturation, temperature. What isn’t clear is whether they can enable these ‘observations’ to be measured feasibly in a range of examples i.e. not just in hospital but at home; not just in half an hour snapshots of time but continuously, not just collecting that data but interpreting it and then becoming assistive (e.g. escalating appropriately)?

More importantly in the paediatric population, can we keep children safer? Could wearables contribute towards the prevention of abuse (for example Sensors within baby grows and sensors to detect smoke from drugs), prevent trafficking with trackers, cot death (such as detection of oxygen saturation) and preventable deaths such as drowning (detecting submergence in water). Wearables do assist with home monitoring such as sleep studies, heart rate monitors, glucose tracking and activity levels and wearables can remind about the timing of medication, rehabilitation exercises or mindfulness. However these functions in a real-time for children and young people have not been proven to improve outcomes.

The breadth of possibilities is large. Virtual reality wearables are better known for their gaming potential but potential future applications in healthcare could include post-operative or post critical care rehabilitation. Another possible application is their use for phobia treatment whereby a virtual environment is used first for a slow and controlled exposure of the patient to whatever it is they are afraid of in order to aid their cognitive training. The most widely implement digital health wearables in paediatrics at this point in time are likely to be the ones used in the diabetic population.

Limitations and obstacles to use

Wearables are currently underused in healthcare,[73] there are layers of governmental and science networks committed to innovation, and many health wearables being developed, but there is a missing layer of implementation, education and training and therefore adoption. The introduction of wearables needs to be set up with appropriate training, education, oversight, departmental leads, key performance indicators (KPI), clinical evaluation, audits and the patient voice. Unless there is funding, development, education, training and time dedicated to NHS workers in why they should trust wearables, how to risk-evaluate, and when to effectively adopt wearables into a clinical pathway, the value in changing lives by 2040 will be lost.

The danger of large amounts of data from wearables is that it must be collected safely, with the patient’s interest at heart, from businesses with NHS values including a strong ethical and moral code. Wearables make the future look exciting and full of promise, but the short-term gain needs to be weighed up against long term risks, and every clinician needs to be as literate in weighing up those risks as they are at critically evaluating a research paper.

Wearable devices to monitor or detect medical conditions are not new to the NHS but never in this widespread manner. NHS technology of the past has never been encumbered with vocabulary such as ‘cool’ or ‘fashionable’. But here is an opportunity to marry up with popular culture, where members of the public will willingly buy and wear devices to provide data norms or provide real time information about medical conditions. While this is a strategy the NHS could embrace this immediately blurs boundaries about regulation: Should they be carefully regulated medical devices with safe data collection and trialled outcomes? With that comes expensive medical CE marks and arduous quality management systems that will immediately decrease the availability and increase the price of the product and create barriers to uptake. Or should they be a non-regulated product that is affordable and allows the public to steer what they need and want and allow systems and medicine to learn from that data.

Lack of peer review and concerns about data bias and safe data collection, (as well as questions about who owns the data) applies to both Non- regulated wearables and even regulated medical devices. Regulation is often unclear and digital regulation policies are newly emerging or incomplete at present as governing bodies try to catch up with digital solutions. On top of this in 2020 the UK is leaving Europe just as Europe have updated their Medical Device Directive (MDD) to the more stringent better- regulated MDR (Medical Device Regulations). An individual digital ethical and moral understanding is needed which must involve the voices of children and young people.

Case study: networks that grow innovation

The National TITCH Network / NIHR CYP MedTech experience

In 2014, the first national network to support the development of child health technology was developed – TITCH (Technology Innovation Transforming Child Health). This was driven by a clear need to bring centres together to work collaboratively on the development of technology that was specific to children and young people, thereby providing a scalable offering to industry and thus overcoming the perception that the paediatric healthcare market is too small to invest in. The development of technology for children and young people at this time was, and still remains a localised and fragmented process with limited spread of impactful technologies beyond the confines of Trusts or at best regions. Child health technology development often resulted from repurposing of adult technologies with suboptimal results or worse still new and challenging complications. Furthermore, technology developed for children lacked the versatility to address the significant anatomical, physiological and developmental changes in childhood that would require technology to adapt or be adapted to meet the needs of the paediatric population.

The National TITCH Network brought together clinicians, academics, designers, computer scientists, private sector partners from large and small multinationals and most importantly children and their parents to identify unmet needs, support collaborations, support the development and evaluation of child health technology. Through a series of workshops, the TITCH Network identified over 100 unmet needs. By working with funders, the TITCH network leveraged over £6 million in funding for technology projects that were specific to children and young people. The TITCH network has grown substantially over the last few years, and now comprises over 40 members with expertise in a broad range of professions, including paediatric medicine, paediatric surgery, community medicine, general practice, allied health professions, small and medium sized enterprises, large corporations, academia, design, engineering, education, and patient/parent/carer representation.

One of the important aspects of the TITCH work was to identify the barriers to the development and implementation of child health technology, and to understand the way in which organisations across the private and public sector could work collaboratively to accelerate digital and device development. Published in 2017 and supported by the Northern Health Science Alliance (NHSA), the TITCH Network published a document that focused on supporting life course technology development as a clinical industry partnership.[74] The paper arose from workshops bringing together delegates including academics, clinicians, funding body representatives, industry partners, colleagues from the NHS, NIHR and AHSNs as well as national regulatory bodies. A series of recommendations were published in this document, founded on key findings.

One important finding that resonates throughout the work focusing on the development of child health technology is the need to include children and young people. There is a need to ensure that the service is ready or can adapt to new technologies, and that there is user-acceptance. There is clear conflict between what works and what works for the patient. While children’s healthcare has some excellent examples of technology that has provided benefit to children with health needs, a lot of technology has been rejected because the technology was functionally inappropriate for patients or that it could not be consolidated into the normal daily living of children and young people. As we move forward it is vital that we develop opportunities and new methodologies to support the development and real-world evaluation of health technologies for children and young people at pace, and drive spread to ensure that the benefit is far-reaching.

As the TITCH Network grew, there was a clear need to support and focus on technology development in key clinical areas where unmet needs had been identified. This resulted in the development and funding of the first paediatric NIHR MedTech and In-vitro diagnostic Cooperative (MIC) namely the NIHR Children and Young People MedTech Cooperative (NIHR CYP MedTech). NIHR CYP MedTech works clinical teams, businesses, academics, charities, children, young people, and families to support the development and adoption of technology for child health and paediatrics. NIHR CYP MedTech is hosted by Sheffield Children’s NHS Foundation Trust and focuses on early stage technology development and prioritises innovations that fit within seven clinical themes. Each theme is led by one or two leading clinicians based at various NHS Trusts across England.

NIHR Children and Young People MedTech Co-operative is one of the 11 MedTech and In vitro diagnostics Co-operatives (MICs) funded by the NIHR. The aim of the 11 MICs is to work with clinical teams, academics, businesses, and service users to support the development and adoption of new medical devices, healthcare technologies, and technology-dependent interventions in the NHS. The NIHR CYP MedTech 2018-2019 biennial report published last year demonstrated the value of working as a collaborative network to support technology development for children and young people in a number of clinical areas.[75]

In the first 2 years NIHR CYP MedTech has leveraged over £6 million in funding for child health technology development, identified 114 unmet needs, worked with 71 SMEs and 23 global organisations, and worked on 60 projects, 40% of which had an industry partner or input. One of the key goals for NIHR CYP MedTech in the future is to ensure that we drive the spread of health technologies to ensure that there is equitable access to advanced therapies for children and young people.


VisualV1 - Plus Created with Sketch. VisualV1 - Plus Copy Created with Sketch.

Benniche, S.; Firth, A.; Jacob, H.; Roland, D. The Impact of Health Technology on the Delivery of Paediatric Care: A Narrative Review. Preprints 2021, 2021020462 (doi: 10.20944/preprints202102.0462.v1).


Mousavi Jazayeri, S. and Jamshidnezhad, A. (2019). Top Mobile Applications in Pediatrics and Children’s Health: Assessment and Intelligent Analysis Tools for a Systematic Investigation. Malaysian Journal of Medical Sciences, 26(1), pp.5-14.


Weekly, T., Walker, N., Beck, J., Akers, S. and Weaver, M. (2018). A Review of Apps for Calming, Relaxation, and Mindfulness Interventions for Pediatric Palliative Care Patients. Children, 5(2), p.16.


NHS Apps Library.


Mousavi Jazayeri, S. and Jamshidnezhad, A. (2019). Top Mobile Applications in Pediatrics and Children’s Health: Assessment and Intelligent Analysis Tools for a Systematic Investigation. Malaysian Journal of Medical Sciences, 26(1), pp.5-14.


Crico, C., Renzi, C., Graf, N., Buyx, A., Kondylakis, H., Koumakis, L. and Pravettoni, G. (2018). mHealth and telemedicine apps: in search of a common regulation. ecancermedicalscience, 12.


Crico, C., Renzi, C., Graf, N., Buyx, A., Kondylakis, H., Koumakis, L. and Pravettoni, G. (2018). mHealth and telemedicine apps: in search of a common regulation. ecancermedicalscience, 12.


Harwich EL, Kate. Thinking on its own: AI in the NHS: Reform Health; 2018


Hall, Pesenti. Growing the artificial intelligence industry in the UK 2017


Topol E. The Topol Review: Preparing the healthcare workforce to deliver the digital future. 2019 2018-02-21.


Grace K, Salvatier J, Dafoe A, Zhang B, Evans O. When Will AI Exceed Human Performance? Evidence from AI Experts. 2017.


Topol E. The Topol Review: Preparing the healthcare workforce to deliver the digital future. 2019 2018-02-21.


Topol E. Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again: Basic Books; 2019.


Topol E. High-performance medicine: the convergence of human and artificial intelligence. Nature Medicine. 2019;25(1):44-56.


Ardila D, Kiraly AP, Bharadwaj S, Choi B, Reicher JJ, Peng L, et al. End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography. Nature Medicine. 2019;25(6):954-61.


McKinney SM, Sieniek M, Godbole V, Godwin J, Antropova N, Ashrafian H, et al. International evaluation of an AI system for breast cancer screening. Nature. 2020;577(7788):89-94.


Mitani A, Huang A, Venugopalan S, Corrado GS, Peng L, Webster DR, et al. Detection of anaemia from retinal fundus images via deep learning. Nature Biomedical Engineering. 2020;4(1):18-27.


De Fauw J, Ledsam JR, Romera-Paredes B, Nikolov S, Tomasev N, Blackwell S, et al. Clinically applicable deep learning for diagnosis and referral in retinal disease. Nature Medicine. 2018;24(9):1342-50.


Google Duplex: A.I. Assistant Calls Local Businesses To Make Appointments 2018.


Ronicke S, Hirsch MC, Türk E, Larionov K, Tientcheu D, Wagner AD. Can a decision support system accelerate rare disease diagnosis? Evaluating the potential impact of Ada DX in a retrospective study. Orphanet Journal of Rare Diseases. 2019;14(1):69.


Botha, J.S.F., Scheffer, C., Lubbe, W.W. et al. Autonomous auscultation of the human heart employing a precordial electro-phonocardiogram and ensemble empirical mode decomposition. Australas Phys Eng Sci Med 33, 171–183 (2010).


BabyTalk: Generating Textual Summaries of Clinical Temporal Data, 2019, University of Aberdeen,


François Portet, Ehud Reiter, Albert Gatt, Jim Hunter, Somayajulu Sripada, Yvonne Freer, Cindy Sykes, Automatic generation of textual summaries from neonatal intensive care data, Artificial Intelligence, Volume 173, Issues 7–8, May 2009, Pages 789-816


Oracle 2020. What Is Big Data?


De Mauro A. A formal definition of Big Data based on its essential features. Library Review. 2016;65(3):122-35.


Pastorino R, Sezione di Igiene IdSPUCdSCRI, Department of Biology UoPPG, De Vito C, Department of Public H, Infectious Diseases SUoRRI, et al. Benefits and challenges of Big Data in healthcare: an overview of the European initiatives. European Journal of Public Health. 2020;29(Supplement_3):23-7.


Big Data: The next frontier for innovation, competition, and productivity.


Wang Y, Kung L, Byrd TA. Big data analytics: Understanding its capabilities and potential benefits for healthcare organizations. Technological Forecasting and Social Change. 2018;126:3-13.


Wang Y, Kung L, Byrd TA. Big data analytics: Understanding its capabilities and potential benefits for healthcare organizations. Technological Forecasting and Social Change. 2018;126:3-13.


Big Data: The next frontier for innovation, competition, and productivity.


Wang Y, Kung L, Byrd TA. Big data analytics: Understanding its capabilities and potential benefits for healthcare organizations. Technological Forecasting and Social Change. 2018;126:3-13.


Raghupathi W, Raghupathi V. Big data analytics in healthcare: promise and potential. Health Inf Sci Syst. 22014.


Godown J, Gaies M, Wilkinson JD. Leveraging big data to advance knowledge in pediatric heart failure and heart transplantation. Tran, 2019,


Sutherland SM. Big Data and Pediatric Acute Kidney Injury: The Promise of Electronic Health Record Systems. Front Pediatr. 2020;7:536. Published 2020 Jan 14. doi:10.3389/fped.2019.00536


Godown J, Gaies M, Wilkinson JD. Leveraging big data to advance knowledge in pediatric heart failure and heart transplantation. Tran, 2019,


Alder Hey Children’s Hospital 2019 [updated 2019-02-11.


IBM and the NHS: Working in Partnership – IBM United Kingdom – Blog. 2019.


Bradford Teaching Hospitals NHSFT. Trust to launch Europeâ™s first AI-powered hospital command centre â“ Bradford Teaching Hospitals NHS Foundation Trust 2018


Business Wire. Bradford, GE Healthcare Announce AI-Powered Hospital Command Center, First of its Kind in Europe | Business Wire: @businesswire; 2018 [updated 2018-10-04.


Harvard Business School. Capacity Command Center: Machine Learning & Hospital Management at Johns Hopkins Technology and Operations Management 2020


Financial Times. Bradford hospital launches AI powered command centre 2018 [updated 2018-10-15.


RCPCH, The National Neonatal Audit Program online, 2019,


Imperial College London, Neonatal Analysis Data Unit, 2019,


Sutherland SM. Big Data and Pediatric Acute Kidney Injury: The Promise of Electronic Health Record Systems. Front Pediatr. 2020;7:536. Published 2020 Jan 14. doi:10.3389/fped.2019.00536


Hodgson LE, Sarnowski A, Roderick PJ, Dimitrov BD, Venn RM, Forni LG. Systematic review of prognostic prediction models for acute kidney injury (AKI) in general hospital populations. BMJ Open. 2017;7(9):e016591. Published 2017 Sep 27. doi:10.1136/bmjopen-2017-016591


Nervecentre. Nottingham University Hospitals NHS Trust become the first UK trust to deploy a whole hospital platform using mobile devices for each of their 4000 clinical staff.


European Parliament. Digitalisation and Big Data: Implications for the health sector 2018


Dhindsa K, Bhandari M, Sonnadara RR. What’s holding up the big data revolution in healthcare? BMJ. 2018:k5357.


European Parliament. Digitalisation and Big Data: Implications for the health sector 2018


Lord Carter, PR. (2018). NHS operational productivity: unwarranted variations. NHS Improvement.


Oxford Dictionary,


NICE Medtech innovation briefing [MIB78]. QuikRead go for C-reactive protein testing in primary care


NICE guidance search, published type=mib


Sim4med website


Kamphuis, Carolien, Esther Barsom, Marlies Schijven, and Noor Christoph. “Augmented reality in medical education?.” Perspectives on medical education 3, no. 4 (2014): 300-311.


Sakakushev, Boris E., Blagoi I. Marinov, Penka P. Stefanova, Stefan St Kostianev, and Evangelos K. Georgiou. “Striving for better medical education: the simulation approach.” Folia medica 59, no. 2 (2017): 123-131.


Riva, G. (2003). Applications of virtual environments in medicine. Methods of Information in Medicine, 42, 524-534.


Beolchi L, Riva G. Virtual reality for health care. In: Akay M, Marsh A, editors. Information Technologies in Medicine. Vol. 2. New York, USA: John Wiley & Sons, Inc.; 2001. pp. 39-83. DOI: 10.1002/0471206458.ch3


Barsom, Esther Z., Maurits Graafland, and Marlies P. Schijven. “Systematic review on the effectiveness of augmented reality applications in medical training.” Surgical endoscopy 30, no. 10 (2016): 4174-4183.


Persson, Johanna. “A review of the design and development processes of simulation for training in healthcare–A technology-centered versus a human-centered perspective.” Applied ergonomics 58 (2017): 314-326.


Hodges, Larry F., Barbara O. Rothbaum, Rob Kooper, Dan Opdyke, Thomas Meyer, M. North, J. J. de Graff, and J. Williford. “Virtual environments for exposure therapy.” IEEE Computer Journal, July (1995): 27-34.


Satava, Richard M. “Virtual reality surgical simulator.” Surgical endoscopy 7, no. 3 (1993): 203-205.


Shannon’s Liu. ‘Wearable technology- statistics & facts’. Technology & Telecommunications. ( March 2019.


Public Health England, Diabetes UK, NHS England NHS Diabetes Prevention Programme (NHSDPP) service specification, overview and FAQ. NHS England Publications Gateway Reference 05728. first published 26 August 2016. updated:23 August 2019


Appelboom, G., Camacho, E., Abraham, M.E. et al. Smart wearable body sensors for patient self-assessment and monitoring. Arch Public Health 72, 28 (2014) doi:10.1186/2049-3258-72-28

VisualV1 - Arrow Created with Sketch. VisualV1 - Arrow Created with Sketch.