Using Computational Video Analysis in Aging Mice to Evaluate the Effects of Chronic Monotherapy, Polypharmacy, and Deprescribing Over Time

BACKGROUND: In clinical studies of older adults, polypharmacy (use of ≥ 5 drugs) and the Drug Burden Index (measures exposure to anticholinergic and sedative drugs) are associated with impaired physical function and frailty. We used computational video analysis of aging mice to examine the impact of medications on morphometric and gait function. METHODS: Middle-aged (12-month) male mice were administered therapeutic doses of medications in polypharmacy regimens with different Drug Burden Index scores or monotherapy with medications from the High Drug Burden Index polypharmacy regimen. At age 21 months, half of the treated animals had their medications deprescribed (discontinued). Open field videos and mouse clinical frailty index were recorded at 12, 15, 18, 21, and 24 months. After applying open-source neural networks to the videos, the gained features were analyzed to detect differences between the treatment groups and control over time using a state-space model with change point detection. RESULTS: We measured 49 morphometric and gait features for 278 mice. The sum of effects of constituent monotherapies did not equal the effects of polypharmacy. Consistent with clinical data, greater gait and posture changes were observed with polypharmacy regimens with increasing Drug Burden Index scores. Deprescribing effects varied between medications, consisting of reversible, irreversible, and novel changes. Different medication exposures had different effects on gait, posture, and the prediction of frailty. CONCLUSION: Computational video analysis of preclinical data is a promising tool for high-throughput, sensitive detection of medication effects in aging.