GUILDFORD — University of Surrey researchers have developed an artificial intelligence tool capable of predicting three distinct long-term health trajectories for heart attack survivors, paving the way for targeted medical interventions immediately following a cardiac event.
The study, published in the Journal of the American Medical Informatics Association, analyzed health records from 12,701 UK Biobank participants who had survived a heart attack. Using data-driven machine learning, the research team mapped the sequence, timing, and biological pathways of secondary diagnoses over a five-year post-event window.
Three Distinct Patient Trajectories
The machine learning model categorized patients into three primary recovery profiles based on their post-incident health progression:
- Cardiometabolic Group (63 percent) – The largest cohort developed conditions such as hypertension, type 2 diabetes, and dyslipidemia, alongside episodic cardiac and respiratory complications.
- Smoking-Related Group (23 percent) – Comprising predominantly smokers, this group experienced multi-organ decline affecting the lungs and musculoskeletal system. This cohort recorded a 44 percent mortality rate—more than triple that of the cardiometabolic group.
- Structural & Kidney Group (14 percent) – This pathway was characterized by structural heart disease, arrhythmias, and renal complications.
Lead author Dr Anthony Onoja noted that a patient’s trajectory could be predicted at the moment of the heart attack using pre-existing diagnoses and demographic data, with older age, respiratory conditions, and higher deprivation scores serving as critical indicators for the highest-risk group.
Biological Pathways and Clinical Application
Genetic analysis confirmed that each health trajectory corresponds to specific molecular mechanisms: immune activation and tissue remodeling in the cardiometabolic group, insulin signaling and lipid transport in the structural group, and chronic inflammation with systemic degeneration in the smoking-related cohort.
Senior author Professor Nophar Geifman emphasized that while traditional tools like the SMART score remain effective for evaluating recurrent cardiac risk, the AI model provides deeper insights into specific disease mechanisms, allowing clinicians to determine precisely where and how to intervene.
