Nurse Stress Prediction Dashboard

Published December 1, 2025

Data Analytics Software QA Janaka Alwis

Introduction

This study presents the development of a Nurse Stress Prediction Dashboard using the Dashtera no-code analytics platform. The system visualizes and analyzes physiological signals collected from wearable sensors kto identify stress patterns among nurses. The dataset comprises more than 11 million time-stamped records from 15 participants, including measurements of electrodermal activity (EDA), heart rate (HR), temperature (TEMP), and triaxial accelerometer data (X, Y, Z). Through the integration of Dashtera’s visual analytics capabilities, this project demonstrates a scalable framework for stress detection and behavioral pattern analysis based on physiological and contextual indicators.

The increasing prevalence of occupational stress among healthcare professionals has motivated the use of wearable sensor analytics to monitor physiological responses during work activities. Nurses, in particular, experience high levels of cognitive and physical stress, which can affect both performance and well-being.

To support data-driven assessment of these phenomena, this project introduces the Nurse Stress Prediction Dashboard, an interactive, no-code analytical interface developed using Dashtera. The dashboard enables exploratory analysis of stress-related physiological signals to uncover relationships between electrodermal activity, cardiovascular response, temperature variation, and movement intensity.

By transforming high-frequency sensor data into interpretable visual representations, the system provides an evidence-based approach for studying stress patterns, activity levels, and physiological correlations in nursing practice.

Dataset

The dataset used in this project was obtained from continuous physiological monitoring of 15 nurses using wearable sensor devices. Each record represents a synchronized time sample containing accelerometer, electrodermal, cardiac, and thermal data, along with a categorical stress label.

Variables and Measurements

Variable	Description
X, Y, Z	Triaxial accelerometer readings representing wrist orientation and movement
EDA	Electrodermal Activity (microsiemens), a proxy for sympathetic nervous system arousal
HR	Heart Rate (beats per minute)
TEMP	Skin Temperature (°C)
id	Unique nurse identifier
datetime	Timestamp of observation
label	Stress classification (0 = Calm, 1 = Mild Stress, 2 = High Stress)

The dataset captures multi-day recordings spanning April to June 2020, resulting in a total of 11,988,051 records. The large-scale temporal structure of the dataset allows for fine-grained analysis of stress trends and physiological dynamics at both individual and group levels.

Dataset Description

A summary of the dataset distribution across participants is provided below:

Nurse ID	Record Count
nurse_15	309,131
nurse_5C	865,930
nurse_6B	825,549
nurse_6D	591,363
nurse_7A	1,377,342
nurse_7E	253,447
nurse_83	1,372,819
nurse_8B	423,763
nurse_94	586,096
nurse_BG	608,655
nurse_CE	814,087
nurse_DF	908,168
nurse_E4	1,487,871
nurse_EG	549,124
nurse_F5	535,706
Total	11,988,051

This distribution highlights the varying recording durations and data volumes across participants, which may correspond to differences in shift length, device uptime, or sampling continuity.

About Dashtera

Dashtera is a cloud-based, no-code analytics platform designed to support the visual exploration and analysis of complex datasets. The platform enables users to construct interactive dashboards without programming, allowing for efficient examination of multidimensional data through line plots, bar charts, maps, regressions, and statistical summaries. Its interface allows data to be filtered, compared, and inspected from multiple perspectives, which makes it suitable for exploratory data analysis tasks.

Key Features

Integration with multiple data sources, including CSV files, APIs, and external repositories.
Support for a wide range of visualization types, such as line charts, bar charts, Pareto charts, and geographic maps.
Interactive drill-down capabilities for detailed examination of specific data segments.
Dynamic filtering that enables focused analysis based on selected criteria.

Built-in options

For this project, Dashtera facilitated the transformation of raw physiological data into interactive visual formats, supporting both macro-level summaries and micro-level signal exploration.

Dashboards

Overview

The Overview Page introduces the dataset structure and global trends. A table summarizes record counts per nurse, providing an overview of data completeness and distribution.

A derived metric, Movement Intensity, quantifies physical motion using triaxial accelerometer data:

Movement Intensity = √(X² + Y² + Z²)

This measure reflects the magnitude of motion at each timestamp, enabling differentiation between physical activity and rest. When analyzed alongside stress labels, it supports the identification of emotionally induced stress versus activity-driven physiological variation.

Visualization Components

Line Charts: Display minute-level averages for Electrodermal Activity (EDA), Heart Rate (HR), and Temperature (TEMP).
Stacked Bar Chart: Illustrates daily stress-level distributions (Calm, Mild, High) for Nurse 5C across multiple observation dates.
Point Chart: Depicts Movement Intensity over time for 23 June 2020, highlighting periods of activity and stillness.

Collectively, these visualizations provide an initial overview of nurse-specific physiological states and their temporal progression.

Physiological Signals – Nurse 5C

This section focuses on the physiological signal analysis of Nurse 5C, who contributed one of the largest and most continuous datasets in the study, totaling approximately 865,930 records. The analysis centers on identifying temporal trends and correlations among the principal physiological parameters-Electrodermal Activity (EDA), Heart Rate (HR), and Skin Temperature (TEMP)-across multiple observation days.

The data for Nurse 5C were recorded over five distinct days: 14th April 2020, 8th May 2020, 23rd June 2020, 24th June 2020, and 25th June 2020. Each record contains minute-level temporal granularity, allowing for detailed examination of short-term physiological variations throughout the nurse’s work shifts.

Each parameter is visualized as a continuous signal over time, revealing fluctuations associated with daily work activities and potential stress episodes. This level of granularity allows for temporal segmentation and cross-comparison between physiological markers.

Detailed Analysis – 23rd June 2020 (Nurse 5C)

The final dashboard page provides an in-depth examination of 23 June 2020, integrating spatial motion data with physiological variables.

Visual Components

3-Axis Motion Line Chart: Illustrates simultaneous variation in X, Y, and Z accelerometer axes.
3D Motion Plot: Provides a spatial representation of wrist orientation and movement.
Motion Intensity Time Series: Quantifies total motion amplitude across time.

Correlation Analyses

Motion Intensity vs EDA, HR, and TEMP: Scatter plots showing physiological responses relative to physical movement.
EDA vs HR, EDA vs TEMP, HR vs TEMP: Bivariate plots examining interdependence among physiological indicators.

These visual analyses enable identification of co-activation patterns, such as simultaneous peaks in EDA and HR without corresponding motion, indicative of emotional rather than physical stress responses.

Discussion

The Nurse Stress Prediction Dashboard demonstrates the feasibility of integrating multimodal physiological data into a unified analytical environment using a no-code platform. The analysis highlights several key observations:

Electrodermal Activity (EDA) and Heart Rate (HR) exhibit strong temporal correlations during stress-labeled periods.
Movement Intensity provides contextual information, distinguishing between motion-induced and psychological stress.
Temperature (TEMP) exhibits slower variations, serving as a stabilizing parameter for contextual interpretation.
Day-level segmentation facilitates the identification of periodic stress cycles corresponding to shift structures.

These findings align with existing literature on psychophysiological stress detection, supporting the use of EDA and HR as reliable indicators of autonomic arousal.

Conclusion

This study demonstrates the application of Dashtera as a no-code analytical framework for stress prediction and physiological data visualization. By consolidating accelerometer, electrodermal, cardiac, and temperature data into a coherent dashboard structure, the system enables real-time exploratory analysis of occupational stress among nurses.

The approach underscores the value of combining wearable sensor data with interactive visualization tools to support health monitoring and workplace well-being. Future extensions of this work may involve predictive modeling, anomaly detection, and cross-subject pattern generalization using the same data infrastructure.