Konstantin Georgiev, Health Data Scientist at Red Star, delivered a 10-minute presentation at the European Society for Geriatric Medicine (EuGMS) conference, showcasing Red Star’s innovative dementia modelling work. The session received positive feedback from professionals in geriatric care, marking another significant milestone for our team.
Our study aimed to evaluate the potential of a machine learning (ML) model for predicting dementia up to 13 years before diagnosis. Using a linked routine primary and secondary care database, we focused on a large, unrestricted cohort of 144,000 Scottish older adults registered in NHS Lothian between 2009 and 2023. Training data was collected in the first year, with follow-up extending up to 13 years. During this period, 8% of participants developed dementia, identified through HDRUK CALIBER coded phenotypes. The model utilized 171 routine health variables, including patient background, lifestyle risk factors, lab results, prescribing history, comorbidities, frailty markers, and routine spirometry and blood pressure data. Of these, 22 variables were selected with geriatrician input to compare against a clinically-supervised model
Our abstract will be included in the EuGMS conference proceedings, where Red Star will also be mentioned. Stay tuned for further updates as we continue to advance our work in dementia risk stratification and early intervention!