從智慧型手機的數位足跡解析人類日常行為：以手機使用、睡眠作息和工時型態為例

"Interpretation of Daily Human Behavior via Smartphone Digital Footprints: Examples from Smartphone Use, Sleeping Patterns, and Working Hours"

近年來普及全球的智慧型手機，使我們可以透過每天人和手機互動留下的「數位足跡」，更準確、即時、持續地量化日常環境中的個人心理與行為。研究手機的數位足跡將直接從人類行為紀錄的巨量資料獲取珍貴的訊息，催生出「網路心理學」、「心理資訊學」、「數位表現型」等新興研究領域，這是突破傳統研究模式的新進展。本文將回顧數個以智慧型手機被動資料為基礎的行為模式研究，介紹如何收集並分析手機資料以了解人類一天常見的三項行為：使用手機、睡眠與工作。從這些實證應用中，本文綜合整理了手機資料較傳統研究方法的幾項重要優勢：提高受試者留存率，避免時間扭曲效應，以及提升資料的時間解析度。文末並討論手機數位足跡未來發展的潛力與優勢：除了解讀人類心理與行為以外，還可以進行即時、密集、個人化的行為介入與治療。

"With the global prevalence of smartphones in recent years, we can now study human behavior and the mind via our daily interactions with smartphones. These data automatically collected by our digital devices are called “digital footprints.” They not only provide an objec-tive, real-time, and ecological source of measurement, but also provide insights into human behaviors and mental activities. The digital foot-prints from smartphones can be seen as a new opportunity for behav-ioral science and psychological research, for example, in the emerging fields of cyberpsychology, psychoinformatics, and digital phenotyping. This review introduces several studies that have applied time-series smartphone passive data to interpret common human behaviors, focus-ing on three mobile apps: “Know Addiction” for smartphone use, “Rhythm” for sleep time, and “Staff Hours” for working hours. “Know Addiction” automatically records the timestamps of screen-on, screen-off, notifications, and app usage. First, we defined an ‘epi-sode’ of smartphone use as the time period from screen-on to the suc-cessive screen-off. App-generated parameters reflecting the frequency and duration of smartphone use facilitate the identification of smart-phone addiction. Second, we shifted from smartphone-centered analysis to human-centered analysis by distinguishing “proactive use” from “reactive use.” Our prior research has shown that the duration of pro-active use, defined as the total time of the epochs without any notifica-tion within one minute before the screen-on, may be more representative of addictive behavior than the total duration of smartphone use. Third, by applying methods like empirical mode decomposition to identify trends in smartphone use, we are able to observe long-term behavioral patterns. “Rhythm” was designed to identify sleep time based on smart-phone behaviors. “Rhythm” also measures changes in sleep patterns and promotes users’ awareness of social jetlag between weekdays and weekends. By quantifying long-term circadian rhythm stability, a “digital chronotype” can be delineated. Our previous study has shown that screen time, mainly mediated by bedtime smartphone use, delayed the circadian rhythm, and reduced total sleep time. “Staff Hours” is an app to capture working hours and patterns for medical staff in real-time. This app collects objective GPS loca-tion data longitudinally in the background with a power-saving design. Using geofencing technology, combined with self-reported work time information and on-call schedule, this app automatically records the working hours one spends in his or her workplace. “Staff Hours” improves the efficiency of labor inspection, as we can now compare real-time work hours on a large scale. Our prior study revealed that medical staff had longer work hours than non-health-care professionals, with resident physicians working the longest hours at 60.4 hours per week in hospitals. There are several advantages of using digital footprints from smart-phones in behavioral science and psychological research. First, passive data collection solves the problem of recall bias and time distortion, and results in higher user retention and temporal resolution. Moreover, smartphones show potential for immediate interventions and person-alized treatments. With the growing emphasis on medical device soft-ware nowadays, we envision that mobile apps collecting digital foot-prints will be widely used in clinical settings and public health."