AI Identifies Four Recovery Profiles in Older Patients
Researchers from the Information Processing and Telecommunications Center (IPTC) at the Universidad Politécnica de Madrid (UPM), in collaboration with researchers from the Universidad Complutense de Madrid (UCM) and Hospital Universitario Cruz Roja, have applied unsupervised machine learning to better understand how older patients recover during geriatric rehabilitation.
The study, part of the AIRCOT (Artificial Intelligence Red Cross Occupational Therapy) project, analyzed the functional trajectories of 957 older adults admitted to a geriatric rehabilitation unit between 2019 and 2025. Using longitudinal information from the Modified Barthel Index, the researchers applied clustering techniques to identify groups of patients following similar patterns of functional decline and recovery. The robustness of the results was assessed using bootstrap sampling, hierarchical clustering and Gaussian Mixture Models.
The analysis revealed four main recovery profiles: Functional Resilience, High Recovery, Chronic Impairment and Severe Damage. The first two groups showed the most favorable recovery trajectories, while the other profiles experienced more persistent limitations. The study also highlighted that improvements in muscle strength do not always translate into equivalent improvements in walking ability, underlining the complexity of functional recovery in older adults.
These results demonstrate how data-driven methods can help identify clinically meaningful patterns within a highly heterogeneous rehabilitation population. In the future, this approach could contribute to more personalized rehabilitation strategies, improved patient stratification and better allocation of rehabilitation resources. It could also provide a foundation for developing predictive models based on patients’ characteristics at admission.
The study represents an example of how artificial intelligence and data science can be combined with clinical expertise to support more individualized approaches to healthy ageing and rehabilitation. The authors emphasize, however, that the identified profiles are currently retrospective and require prospective, multicenter validation before they can be used to guide clinical decisions.
Bibliographic reference:
Martínez-Zujeros, S., Zufiria, P. J., Sasot, A. V., & Delgado-Losada, M. L. Robust Unsupervised Analysis of Longitudinal Functional Trajectories in Older Inpatients Reveals Stable Recovery Profiles: The AIRCOT Study. European Journal of Investigation in Health, Psychology and Education, 16(8), 111. https://doi.org/10.3390/ejihpe16080111
Share this:
