A Longitudinal Health and Lifestyle Dataset of 60 Students for AI-Based Monitoring and Digital Twin Applications

The dataset titled “A Longitudinal Health and Lifestyle Dataset of 60 Students for AI-Based Monitoring and Digital Twin Applications” presents a structured time-series collection of physiological and behavioral data from 60 college-going individuals (age 18-25) over 75 consecutive days. It contains 4500 records (60 users × 75 days) with 14 features representing key health indicators and lifestyle attributes. The dataset captures real-world variability in student routines, including sleep patterns, academic workload, physical activity, and screen exposure, making it suitable for AI-driven health monitoring, behavioral modeling, and digital twin systems.

The recorded attributes include:

-Profile ID: Unique identifier assigned to each participant.
-Date: Day of data recording (YYYY-MM-DD).
-Heart Rate (Morning): Resting heart rate in bpm.
-Heart Rate (Evening): Evening heart rate in bpm.
-Blood Pressure (Morning): Systolic/diastolic (mmHg), e.g., 120/80.
-Blood Pressure (Evening): Systolic/diastolic (mmHg).
-Blood Sugar (Before Fasting): Glucose level in mg/dL.
-Blood Sugar (After Fasting): Glucose level in mg/dL.
-Sleep Hours: Total sleep duration (hours).
-Deep Sleep Hours: Deep sleep duration (hours).
-Steps: Daily step count (integer).
-Calories Burned: Energy expenditure (kcal).
-Screen Time: Device usage (hours/day).
-Daily Working Hours: Academic/work duration (hours).

Data collection was conducted in real-time using wearable fitness devices (smartwatches and fitness bands), smartphone health tracking applications, and validated digital medical devices such as blood pressure monitors and glucometers. Behavioral attributes such as screen time and working hours were recorded through device usage statistics and manual logs, ensuring realistic representation of student lifestyle patterns and daily variability. The longitudinal nature of the dataset enables analysis of temporal trends, circadian variations, and activity-rest cycles, which are critical for predictive modeling and personalized healthcare applications.

To ensure data consistency and interoperability across devices, unit standardization was applied: heart rate (bpm), blood pressure (mmHg), blood sugar (mg/dL), sleep metrics (hours), steps (integer count), calories burned (kcal), screen time (hours/day), and working hours (hours). Preprocessing included timestamp alignment, normalization of device readings, outlier detection and removal, and validation of missing or inconsistent entries to maintain data quality and reliability.

The dataset is provided in .xlsx format, facilitating integration with machine learning frameworks, statistical analysis tools, and AI pipelines. Its structured time-series design and multi-dimensional features make it suitable for applications in health analytics, digital twin modeling, anomaly detection, lifestyle analysis, and predictive healthcare systems.