Smartphone Keystroke Dynamics Predict Chronological Age
Researchers analyzed free-living smartphone typing logs from 177 healthy adults in South Korea collected between September 2022 and September 2023, totaling over 2.5 million typing sessions. They extracted 43 features at 6-hour, daily, and weekly resolutions and trained eight AI models, with an LSTM achieving MAE 3.69 years (R2=0.71), improving to MAE 3.60 with a custom loss. Findings show age-related temporal keystroke patterns, notably early morning and late evening.
Key Points
- 1Collected 2.5 million typing sessions from 177 adults over approximately 25 weeks each
- 2Showed LSTM model estimated age with MAE 3.69 years and R2 0.71
- 3Indicated passive keystroke features enable unobtrusive age-sensitive monitoring and personalization applications
Scoring Rationale
Strong empirical results and actionable modeling with temporal resolution; limited by relatively small, single-country participant cohort.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
