Advancing machine learning to achieve real-world early detection and personalised disease outcome prediction of inflammatory arthritis
Lead Research Organisation:
University of Reading
Department Name: Business Informatics, Systems and Accoun
Abstract
Over 20 million people in the UK live with rheumatic and musculoskeletal diseases (RMD), and inflammatory arthritis (IA) is a major subdivision of RMD causing joint inflammation leading to damage. IA causes long-term pain, disability and incurs substantial personal and societal costs. There is also an estimated 59% increase in diagnosed IA cases between 2004 and 2020 in the UK which has important implications for health services. Rheumatology departments accounted for approximately 9% of the average NHS trusts total medication spend in 2019/2020. There are still significant unmet needs in the IA patient pathway, especially in IA detection and flare management. IA presents with non-specific symptoms and there is currently no diagnostically definitive single biomarker for IA. Early detection is critical but challenging, and delay in detection and late referral often result in loss of the window of opportunity when effective treatment should start and delays can lead to disability and associated unemployment. For patients who are diagnosed with IA, IA outcomes and activities such as flare-up are very heterogeneous in their manifestations between individual patients. Real-world data from The National Early Inflammatory Arthritis Audit showed inequality in care for rheumatology patients from minority ethnic groups. A lower proportion of ethnic minority patients achieved disease remission compared to white patients. UN4 Finally, weather is another contributing factor of IA flare heterogeneity.
Despite significant unmet needs, RMD, especially IA, is still an underexplored area of real-world ML application in comparison with other diseases. Existing ML studies do not fit for purpose of early detection in practice as they are not trained based on the data available at the point of early detection. Furthermore, although there are studies showing potential determinants of IA, there is no research, or any machine learning methods that can identify the undetected determinants-combination that can offer a useful level of prediction of IA. This is because current ML approaches still cannot handle the underlying relationships among heterogenous datasets with different data types, modalities, contexts, cohorts and levels of incompleteness. On the other hand, existing ML methods in IA, and healthcare in general, still rely on a "one-size-fits-all" paradigm rendering generic learning algorithms, suboptimal on the individual level especially as IA is known to be heterogenous in nature from the time of diagnosis. Although there are methods for explainable ML local, there is limited research to quantify and explain model prediction uncertainty and its usability in practice. For a physician to use and trust ML predictions it is critical to understand the uncertainty associated with these predictions for the individual patient. Although successful translation requires bringing together expertise and stakeholders from many disciplines, the development of ML solutions is currently occurring in silos, and there is a lack of holistic and scalable ML development pipeline. Despite all the limitations of current ML, there are huge opportunities to advance ML, especially in rheumatology applications, because rheumatology has already been leading the way in the use of virtual clinics and remote monitoring in the UK. It is now time to advance ML using data generated for real early detection and personalised management of IA.
Our vision: The proposed project will develop useful and responsible machine learning methods to achieve real-world early detection and personalised disease outcome prediction of inflammatory arthritis. We will develop a holistic and scalable approach through an interdisciplinary team addressing the pressing healthcare challenges of inflammatory arthritis and the limitations of machine learning to accelerate real-world ML application in healthcare.
Despite significant unmet needs, RMD, especially IA, is still an underexplored area of real-world ML application in comparison with other diseases. Existing ML studies do not fit for purpose of early detection in practice as they are not trained based on the data available at the point of early detection. Furthermore, although there are studies showing potential determinants of IA, there is no research, or any machine learning methods that can identify the undetected determinants-combination that can offer a useful level of prediction of IA. This is because current ML approaches still cannot handle the underlying relationships among heterogenous datasets with different data types, modalities, contexts, cohorts and levels of incompleteness. On the other hand, existing ML methods in IA, and healthcare in general, still rely on a "one-size-fits-all" paradigm rendering generic learning algorithms, suboptimal on the individual level especially as IA is known to be heterogenous in nature from the time of diagnosis. Although there are methods for explainable ML local, there is limited research to quantify and explain model prediction uncertainty and its usability in practice. For a physician to use and trust ML predictions it is critical to understand the uncertainty associated with these predictions for the individual patient. Although successful translation requires bringing together expertise and stakeholders from many disciplines, the development of ML solutions is currently occurring in silos, and there is a lack of holistic and scalable ML development pipeline. Despite all the limitations of current ML, there are huge opportunities to advance ML, especially in rheumatology applications, because rheumatology has already been leading the way in the use of virtual clinics and remote monitoring in the UK. It is now time to advance ML using data generated for real early detection and personalised management of IA.
Our vision: The proposed project will develop useful and responsible machine learning methods to achieve real-world early detection and personalised disease outcome prediction of inflammatory arthritis. We will develop a holistic and scalable approach through an interdisciplinary team addressing the pressing healthcare challenges of inflammatory arthritis and the limitations of machine learning to accelerate real-world ML application in healthcare.
Publications
Chan A
(2023)
A concept for digital transformation for improved patient care in the UK
in Health Policy and Technology
Collins MP
(2024)
Vasculitic neuropathy-related disability, pain, quality of life, and autonomic symptoms: A survey of 312 patients.
in Rheumatology (Oxford, England)
Dubey S
(2024)
Artificial intelligence and machine learning in rheumatology
in Rheumatology
Hellmich B
(2024)
EULAR recommendations for the management of ANCA-associated vasculitis: 2022 update.
in Annals of the rheumatic diseases
Laura C Coates
(2024)
Enhancing current guidance for psoriatic arthritis and its comorbidities: recommendations from an expert consensus panel.
in Rheumatology
Wang B
(2023)
Improving triaging from primary care into secondary care using heterogeneous data-driven hybrid machine learning
in Decision Support Systems
Description | The experimental AI model to detect Inflammatory Arthritis using multimodal electronic patient records has been developed. The model can detect and differentiate inflamatory arthritis using data (GP referral letter, clincial information summary and avaiable blood test results) available at referral stage. Current sensitivity and specificity of the model is around 90% which is significsntly higher than human GP and clinicans. Wider evaluation is needed for model robustness and fairness. |
Exploitation Route | This funding outcomes will be used by GP and secondary care clinicians to improve referrals so that patients can get access to the right treatment faster. |
Sectors | Healthcare |
Description | The finding will be piloted in two NHS trusts. Current work has raised the awareness of developing and deploying large language models in local NHS. |
Sector | Healthcare |
Description | Data Institute to support local real-world AI and analytics in NHS |
Geographic Reach | Local/Municipal/Regional |
Policy Influence Type | Contribution to new or improved professional practice |
Impact | Raised awareness of benefits, risk, challenges and methodologies of implementing AI in healthcare in NHS. |
Description | MSc training course focusing on bridging the implementation gap of AI in healthcare |
Geographic Reach | Local/Municipal/Regional |
Policy Influence Type | Influenced training of practitioners or researchers |
Impact | Raised awareness of the public about the importance and barriers to deploying AI in real-world healthcare and the roadmap to overcome those barriers. |
URL | https://www.reading.ac.uk/module/document.aspx?modP=INMR96&modYR=2324 |
Description | Referral Inequalities in Inflammatory Arthritis |
Geographic Reach | Local/Municipal/Regional |
Policy Influence Type | Contribution to new or improved professional practice |
URL | https://q.health.org.uk/idea/2024/referral-inequalities/ |
Description | Machine learning based rheumatic and musculoskeletal disease (RMD) risk stratification to improve disease detection and referral triage in rheumatology |
Amount | £39,991 (GBP) |
Funding ID | NIHR205854 |
Organisation | National Institute for Health Research |
Sector | Public |
Country | United Kingdom |
Start | 06/2023 |
End | 11/2023 |
Description | RMD-Health: Machine learning-enabled decision support system to improve early detection and referral of rheumatic and musculoskeletal diseases |
Amount | £1,165,145 (GBP) |
Funding ID | NIHR206473 |
Organisation | National Institute for Health Research |
Sector | Public |
Country | United Kingdom |
Start | 09/2024 |
End | 10/2027 |
Title | A heterogeneous data-driven hybrid machine learning method to improve triaging from primary care into secondary care |
Description | 1. Research dataset: The dataset is collected from the Rheumatology Department of a large secondary care hospital in the UK. This dataset includes several kinds of data modalities, including the GP referral letters, clincial information summary and the blood test results provided by the GP clinics. 2. Model: We developed a heterogeneous data-driven hybrid machine learning model to handel the different data modalities, including a BERT-based NLP model to handle the free text, a series of classical machine learning methods to handle the blood test results, and a probabilistic model to fuse the predictions of different models. |
Type Of Material | Data analysis technique |
Year Produced | 2022 |
Provided To Others? | No |
Impact | 1. The proposed heterogeneous data-driven hybrid machine learning model achieved a precision of 0.83, recall of 0.82, F1-Score of 0.83, accuracy of 0.82, AUC of 0.90 in identifying patients with non-inflammatory conditions (NIC) and inflammatory arthritis (IA) at the point of triage with explainable risk stratifications. 2. Our model is piloted in a real-world trial in a large secondary care hospital in the UK to compare referral accuracy and time saved between our model and clinicians, and evaluate its acceptability by users. 3. Our model achieved precision and recall of 0.83 and 0.81, compared with the precision and recall of 0.80 and 0.78 by clinicians. 4. The research also shows that our model enabled decision support can save clinicians 8 hours per week in assessing the referral assessment. 5. This research is the first study to streamline hospital triage from primary care to secondary care using machine learning. |
URL | https://www.sciencedirect.com/science/article/pii/S0167923622001701 |
Title | Inflamatory Arthritis flare prediction data and model |
Description | Data of longitudinal blood monitoring and electronic patient-reported outcomes of patients with inflammatory arthritis (IA) at Royal Berkshire NHS Foundation Trust has been collected |
Type Of Material | Database/Collection of data |
Year Produced | 2024 |
Provided To Others? | No |
Impact | This dataset will lead to the development of the IA flare prediction model |
Description | AI application to support primary care referrals |
Organisation | Buckinghamshire, Oxfordshire and Berkshire West Integrated Care Board (BOB ICB) |
Country | United Kingdom |
Sector | Public |
PI Contribution | AI model to detect inflammatory arthritis |
Collaborator Contribution | Future application or our AI model to detect inflammatory arthritis in primary care referrals |
Impact | The model is designed and developed to be fit-for-purpose for both primary care and secondary care. |
Start Year | 2023 |
Description | Personalised analytics using longitudinal blood monitoring data and AI and machine learning-based testing analysis |
Organisation | University of Warwick |
Department | Department of Computer Science |
Country | United Kingdom |
Sector | Academic/University |
PI Contribution | We provide feasibility study funding for the project "Personalised analytics using longitudinal blood monitoring data and AI and machine learning-based testing analysis" |
Collaborator Contribution | We will codevelop the project on "Personalised analytics using longitudinal blood monitoring data and AI and machine learning-based testing analysis" |
Impact | Multi-disciplinary. Outcomes including publication, software and models will be developed. |
Start Year | 2023 |
Title | E-triage to support referral management for elective care |
Description | The e-triage application is developed by Royal Berkshire NHS Foundation Trust to streamline referral and triage process. It will be connected with AI model that detect inflamatory arthritis to improve the referrals in rheumatology. |
Type Of Technology | Software |
Year Produced | 2024 |
Impact | The referral processes at Royal Berkshire NHS Foundation Trust have been improved. |
Title | Risk stratification model for rheumatic and musculoskeletal diseases |
Description | The software package of the AI model to detect and differentiate inflammatory arthritis and non-inflammatory conditions. |
Type Of Technology | Software |
Year Produced | 2023 |
Impact | The impact is being developed and if it is applied in the NHS it will improve referrals in rheumatology. |
Title | Simplifying Machine Learning Library (SMLL) |
Description | SMLL is a Python package featured with SQLiteDB, multithreading, and C/C++ extension support, and aims to speed up the development, training, analysis, as well as deployment of machine learning models used for the AI4IA project. |
Type Of Technology | Software |
Year Produced | 2024 |
Impact | By using SMLL, the development, training, analysis, and deployment of machine learning models could be easier. Besides, SMLL also contains a series of data preprocessing methods and a lot of predefined models that could be used for the AI4IA project. |
Description | Alan Turing Institute workshop: |
Form Of Engagement Activity | A formal working group, expert panel or dialogue |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Professional Practitioners |
Results and Impact | 'Clinicians in the AI loop: exploring decision support tools', jointly hosted by the Turing-Roche partnership and the Turing Clinical AI Interest Group. There is growing interest and research in developing clinical decision support tools that may be able to lessen the burden on our busy healthcare systems. However there are many unanswered questions around these tools, such as how best to incorporate the human aspect within them (at the development, trial and post-market surveillance stages) and a considered approach of their risks and benefits. This multidisciplinary workshop aims to bring together a wide range of stakeholders such as clinicians, patients, data scientists, ethicists, engineers and behavioural scientists to gather an unique and inclusive perspective on these tools. |
Year(s) Of Engagement Activity | 2023 |
URL | https://www.turing.ac.uk/research/interest-groups/clinical-ai |
Description | Digital Health Conference at Stanford University |
Form Of Engagement Activity | A formal working group, expert panel or dialogue |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Public/other audiences |
Results and Impact | Weizi Li talked about how to bridge the implementation gap of AI in healthcare. More than 300 academics, industry, medical and postgraduate students attended the conference and the pannel. |
Year(s) Of Engagement Activity | 2024 |
URL | https://www.timeshighered-events.com/digital-health-2024/agenda/session/1196680 |
Description | HICSS 57 mini-track: Digitally-enabled Blood Testing in Healthcare |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Public/other audiences |
Results and Impact | A mini-track Digitally-enabled Blood Testing in Healthcare was organised at The 57th Hawaii International Conference on System Sciences (HICSS), which sparked discussions and follow-up collaborations with international consortiums. |
Year(s) Of Engagement Activity | 2024 |
URL | https://scholarspace.manoa.hawaii.edu/collections/dee1126f-249b-4895-a070-145447c866e8 |
Description | HICSS-58: mini track "Applying Digital Technologies and AI In Virtual Hospitals: Exploring Global Innovative Models Minitrack" |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Public/other audiences |
Results and Impact | This minitrack showcased different virtual hospital models being implemented in various health systems in different countries, exploring their designs, implementation, evaluation, strengths, and issues needing calibration or attention. |
Year(s) Of Engagement Activity | 2024 |
URL | https://hicss.hawaii.edu/tracks-58/information-technology-in-healthcare/#applying-digital-technologi... |