Mental Health and Cognitive Predictors of Functional Decline in Nursing Home Residents with Type 2 Diabetes

Objectives To identify multidomain predictors of functional decline among nursing home residents with type 2 diabetes (T2D). Methods A secondary analysis of longitudinal Minimum Data Set 3.0 data from Iowa nursing home residents with T2D was conducted. Three machine learning models were developed to predict functional decline, defined as worsening Activities of Daily Living Long-Form scores between first and last assessments. Predictor variables were guided by the Functional Consequences Theory and represented b…

By the Truvace deskOctober 15, 20262 min read1 sourcePeer-reviewedTRV-2026-1473· cite & verify

In brief

Researchers analyzed longitudinal Minimum Data Set 3.0 records for 5,440 Iowa nursing home residents with type 2 diabetes across 430 facilities. They built three machine learning models to predict worsening Activities of Daily Living scores between a resident's first and last assessments.

The best model separated residents who declined from those who did not with an AUC of .93. The influential predictors spanned biological, psychological and environmental domains, and several are modifiable, so the findings suggest where care teams could intervene. The text reports a secondary analysis of existing records and does not show that acting on these predictors improves outcomes.

Main points

The sample was 5,440 residents with type 2 diabetes from 430 Iowa nursing homes, with a mean age of 81 and 63% female.
Functional decline was defined as worsening Activities of Daily Living Long-Form scores between first and last assessments.
The most influential predictors were lower baseline function, poorer cognition, higher depressive symptoms, older age and more assessments.
Residents' belief in improving function, coexisting chronic diseases and rurality were also prominent predictors.

The gain

Machine learning models trained on routine nursing home assessment data predicted functional decline in residents with type 2 diabetes with high accuracy, pointing to modifiable risk factors for early identification.

The rundown

The study was a secondary analysis of Minimum Data Set 3.0 data. Three machine learning models were developed, and predictor variables were chosen using the Functional Consequences Theory, covering biological, psychological and environmental factors.

The multi-class classification model fit best, with an AUC of .93. Baseline function, cognition and depressive symptoms ranked as the strongest predictors, while a resident's confidence in improving function, chronic disease burden and rural location also stood out.

The authors conclude that decline reflects interrelated vulnerabilities, several modifiable. They suggest routine assessment of function, cognition and depression for early risk identification, and interdisciplinary interventions supporting mental health and confidence, particularly in rural and medically complex populations.