Abstract
Objective Integrating technological innovation in clinical big data from Nine Health Global (NHG) and data science Woubot is a prototype precognitive system for community & wound clinics. Focusing on leg ulcers, Woubot will produce recommendations from several thousand possible treatment combinations. Working with suppliers to the National Wound Care Strategy Programme, the project will create a suite of automated software tools with a user-friendly mobile application designed by doctors and nurses for their own use within the NHS. This will generate a personalised care pathway for each patient via a series of recommendations. TRUST4Health will apply the technology to other diseases.
Methods We undertook a feasibility study to test artificially intelligent software on data from Cegedim Thin and the NHS Community Data set. We combined know how from our A.I. diagnostic system Diagbot co-produced with a Chinese partner for grass roots doctors in China and applied data science techniques creating a new AI prototype system. With a consortium led by the Royal College of Surgeons in Ireland (RCSI) we have applied for Horizon 2022 EU government funding to build on the work in wounds and to apply the methods to 3 other vascular clinical diseases stroke, heart failure and dementia . Woubot will use artificial intelligence (AI) to identify people likely to develop chronic leg wounds and manage their preventative care. In those that already have leg wounds, such as diabetic foot ulcers, the software will help to ensure that evidence of effective treatment is turned into simple steps which are available quickly and easily to front-line staff. Our AI software will rapidly sift through millions of data items in secure NHS facilities. This will enable recommendations to be generated via a mobile app. A suite of software tools will generate a personalised care pathway with a series of recommendations for use in the NHS. Most of this care will be delivered by nurses and other healthcare professionals in clinics and the community. Prescriptions, whether for exercise, other lifestyle changes, medication or dressings, will be individualised for each patient based on their history and biological makeup and linked to the latest clinical evidence. We will also use image software to monitor progress easily and accurately.
Results We built a secure platform hosted by UK Cloud ( Nine Health Community Interest Company is an NHS research data organisation)using wound data sourced from Cegedim Thin and the NHS Community Data set (NHS Humber Foundation Teaching Trust) using patient pseudonymised data sources (which have gone through the double de-identification process). Data was reviewed by a statistical expert to exclude bias and included a national sample from primary care and a local sample from Hull and East Riding where the demographic includes both inner city, city and rural and a diverse range of nationalities including black, ethnic and minority groups aged 19–80. We collated and analysed around 2000 comprehensive patient records of those with hard to heal wounds (diabetic foot ulcer and venous leg ulcer) across a 2-year period. A raft of modifiable predictive factors such as Vitamin B12 levels, the impact of BMI on healing were identified and analysed. Isolating the key measures enabled the prediction of time from developing diabetes to developing a foot ulcer and then the ability to predict time to an amputation. These results if validated by further research such as the Horizon 2022 EU Trustworthy A.I. project referred to above would enable targeted management to prevent these sequelae. We have developed clinical algorithms based on the national wound guidelines produced by the NWCSP for some parts of the patient pathway
e.g., initial assessment including red flags. We now need to validate via clinical trials and automate processes, combining existing data collected by the National Minimum Wound Assessment Data Set, our data sets and others @ NHS digital https://digital.nhs.uk/ HES, CSDS and other international data.
Conclusion Woubot https://fundingawards.nihr.ac.uk/award/AI_AWARD01723 has started to identify people likely to develop chronic leg wounds and suggested predictive factors which may prevent amputation and death. The automated identification of these factors will in the next phase enable management of their preventative care. In those that already have leg wounds, such as diabetic foot ulcers, the software will help to ensure that evidence of effective treatment is turned into simple steps which are available quickly to front-line staff. Dressing analysis (size and type over time) suggests a good proxy measure for wound healing. In the next phase recommendations for personalised care will be generated via a mobile app. The software will generate a personalised care pathway with a series of recommendations for use in the NHS. Most of this care will be delivered by nurses and other healthcare professionals in clinics and the community.
Prescriptions, whether for exercise, other lifestyle changes, medication or dressings, will be individualised for each patient based on their history and biological makeup and linked to the latest clinical evidence. The clinician chooses whether or not to accept the recommendations and records their decision.
Following the above results in the area of hard to heal wounds we shared these with the Royal College of Surgeons in Northern Ireland and an expert consortium of data scientists and clinicians which has led to our submission to develop trustworthy clinical
A.I. tools for front line clinicians in stroke, heart failure and vascular dementia. For the first time if successful we intend to explore the common causal factors across all 4 disease areas and create a unique synthetic/augmented data resource for the UK and Europe.