A GP reviewing a patient with multiple long-term conditions must piece together information from hospital, GP, and social care records, then cross-reference it with clinical guidelines and risk predictions—all within a ten-minute appointment. This matters because structured medication reviews, introduced under the NHS Long Term Plan, are meant to catch dangerous drug interactions and unnecessary prescriptions. But the information needed is scattered across systems that do not talk to each other, and there is little guidance on which patients need priority review. DynAIRx targets three groups with poor outcomes: older people with frailty, people with co-existing mental and physical health problems, and those with complex multimorbidity and potentially harmful drug combinations. If successful, the project will build AIs that automatically assemble and summarise a patient’s full care history, cluster similar patient journeys, overlay risk trajectories, and surface relevant guidelines—all fed directly into the prescribing software GPs already use. The result would be a learning system that helps clinicians spot, in seconds, whose medication needs urgent review, without extra paperwork or training. This could quietly transform how the NHS manages the growing number of people living with multiple long-term conditions.
View original technical description
Research question Can AI-augmented summarisation and clustering of patient journeys, combined with risk prediction and existing clinical guidelines, support medicines optimisation for people living with multiple long-term conditions, when implemented in prescriber feedback systems? Background Structured Medication Reviews (SMRs) were introduced in NHS Primary Care Networks to support delivery of the NHS Long Term Plan for medicines optimisation. However, it is often challenging to gather the information needed for these reviews due to poor integration of health records across providers and there is little guidance on how to identify those patients whose medication most urgently requires review. DynAIRx will address these problems by targeting potentially problematic polypharmacy in multimorbidity, focusing on three key groups with poor outcomes: Older people with frailty. People with co-existing mental and physical health problems. Other people with complex multimorbidity (≥4 long-term conditions), potentially problematic polypharmacy (≥10 regular medications), and/or potentially harmful drug-drug interactions. Aim To develop AIs that assemble longitudinal summaries across disparate care records, clustering them and overlaying them with risk trajectories and discoverable clinical guidelines in automated audit and feedback to prescribers – evaluated in the context of SMRs for our three multimorbidity groups. Methods Six work-packages (WPs) integrate closely: WP1: REQUIREMENTS FOR WELL-INFORMED/IMPLEMENTED MEDICATION REVIEWS. Engaging with diverse stakeholders through workshops, focus groups and one-to-one interviews to: understand current SMRs and how they might be augmented with AI. identify potential barriers and facilitators to uptake and utilisation. iteratively review/refine prototypes (from WP3-5 via WP5), shaping implementation plans. WP2: STRUCTURED CLINICAL DATA AND NARRATIVE PROCESSING. Creating datasets for WP3-5. Structured clinical data will be drawn from integrated care records (GP, hospital, and social care) from English integrated care systems covering ~11m population, supplemented with Natural Language Processing of some unstructured clinical text. WP3: STATISTICAL LEARNING AND CLUSTERING FOR MULTIMORBIDITY PREDICTION. Training AIs to identify patterns of conditions, medications, tests, and clinical contacts preceding adverse events across our three multimorbidity and polypharmacy groups. We will build the patterns into causal inference to identify individuals who might benefit most from a medication review. WP4: COMBINED LONGITUDINAL DATA VISUALISATION FOR MEDICATION REVIEWS. Advancing visualisation of longitudinal summaries of multi-provider care records overlain with risk trajectories, combined with key features from AI-learned patterns/structures (WP3) and clinical guidelines (WP1). WP5: PRESCRIBER FEEDBACK AND LEARNING SYSTEM. Implementing and evaluating AI-augmented multimorbidity information in an existing prescribing audit and feedback system – creating a learning system for medicines optimisation (centred on users: WP1). WP6: PUBLIC-PATIENT-PRACTITIONER CO-PRODUCTION/CO-EVALUATION. Embedding co-production and ethical design throughout WP1-5 by involving practitioners and people with lived experience of multimorbidity in all aspects of researching, developing, implementing, and disseminating DynAIRx. Impact and dissemination DynAIRx will advance delivery of SMRs by overcoming the challenge of prescribers not being able, particularly in multimorbidity, to assemble information from disparate clinical information systems, guidelines, and risk models. Impacts and dissemination will leverage our strong track-record of delivering similar projects across the Northern Health Science Alliance, NIHR Applied Research Collaborations, Health Data Research UK, several NHS Integrated Care Systems and NHS England.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know