Life logging and bio-signal characteristics are typically time-series data collected over time, and extracting meaningful data features from accumulated data is clinically significant. This paper aims to review the principles and advancements of artificial intelligence and analytical techniques related to life logging and bio-signals, and propose future research directions for fall prevention and risk factor prediction in the elderly. A literature search was conducted through Web of Science, Google Scholar, PubMed, and Scopus. Search terms included "feature extraction," "electromyography," and "machine learning," covering studies published between 2020 and 2023. A total of 67 papers met the inclusion criteria: 18 papers from 2023, 18 from 2022, 18 from 2021, and 13 from 2020. Feature extraction methods related to fall prediction in the elderly based on life logging and bio-signals were categorized into (1) statistical and physical features, (2) linear features, and (3) neural network-based features. Data characteristics can be broadly classified into statistical and physical features, linear features, and neural network-based features. Recently, the advancement of wearable sensors has increased the need for large-scale time-series data analysis and clinical research, suggesting that future studies should integrate various clinical indicators, particularly health status, to analyze clinical data comprehensively.