Researchers at Loughborough University have developed an advanced artificial intelligence (AI) model designed to predict how long individuals with learning disabilities may need to remain in hospital. The breakthrough, part of the DECODE project, aims to improve patient care, enhance resource planning, and address healthcare inequalities for those with complex health conditions.
The study, published in Frontiers in Digital Health, identified cancer as the primary reason for hospital admissions among patients with learning disabilities, while epilepsy was the most commonly treated condition during inpatient stays. On average, hospital stays last three days, but those exceeding 129 days are often linked to mental health conditions. Longer hospitalisations are more common in patients over 50, those living in deprived areas, individuals with obesity, or those with multiple long-term health conditions.
By analysing GP and hospital records from over 9,600 patients in Wales, the AI model demonstrated 76% accuracy in predicting whether a patient would experience a prolonged hospital stay. Professor Georgina Cosma, an expert in AI for healthcare, explained that the model evaluates factors such as age, medical history, and lifestyle to provide early predictions, allowing hospitals to improve planning and deliver personalised care.
The findings will support NHS efforts to develop risk assessment tools for clinical decision-making. Dr Satheesh Gangadharan, co-lead of the DECODE project, highlighted the potential to reduce hospital admissions by identifying earlier intervention opportunities and encouraging patient involvement in their care. The next phase of research will apply the model to hospital datasets across England to assess its wider applicability.
