Development of a Computational Algorithm for Fatigue and Sleepiness Prediction and Management: Mitigation of Human Errors in High Complexity Operations.

• Grant number: 24/20802-9
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Start date: February 01, 2026
• End date: October 31, 2026
• Field of knowledge: Interdisciplinary Subjects
• Principal Investigator: Tulio Eduardo Rodrigues
• Grantee: Tulio Eduardo Rodrigues
• Company: SAFE AND OPTIMIZED STANDARDS ASSESSORIA TECNICA E DIAGNOSTICOS LTDA
• CNAE: Desenvolvimento e licenciamento de programas de computador customizáveis ; Pesquisa e desenvolvimento experimental em ciências sociais e humanas
• Associated researchers: André Frazão Helene

Abstract

Introduction: Fatigue and sleepiness are aspects that are continually present and have a relevant impact on cognitive performance, representing potential risks of human errors and accidents or serious incidents in operational environments. This is especially relevant for shift workers in the transport industry, hospital settings, mining, oil and gas industry, security, among others. The identification of routines that are biologically unfavorable to work, due to high sleep deficit, disruption of the circadian cycle and excessive workload, is extremely important for the effective management of operational risks in small, medium and large organizations. The use of control and mitigation systems is fundamental in these scenarios, reducing risks and costs in any human action that involves operational routines of relevant complexity.Objectives: Develop a computational algorithm and interface for the analysis and mitigation of fatigue and sleepiness in complex operational environments, with tailored parameters for each environment and adopting improvements that do not exist in currently available tools. Verify relationships between quantity and quality of sleep and work routines, calculating the probabilities of lapses, such as those observed in psychomotor vigilance tests (PVT), with respective confidence intervals. Study the relationship between workload, fatigue and sleepiness, estimating performance levels with uncertainty ranges.Methodology: The study will include a market survey and two convenience samples, one with 20 to 30 pilots of airplanes and helicopters and the other with 30 to 40 helicopter pilots, without distinction of race, gender, age, function or length of service. All participants will be invited to fill out a questionnaire on a web platform, with questions about social aspects, health, sleep habits and strategies, and work routines. Actigraphy data will be collected for a period of one month for each participant. Each participant will make daily entries about the time they went to bed and got up, where they slept and their self-reported sleep quality. Participants must perform psychomotor vigilance tests (PVT), as well as classify their workload, fatigue levels (Samn-Perelli scale) and sleepiness (Karolinska scale) at pre-established times. The work schedules of all participants will also be analyzed. The research project will be submitted for consideration by the Human Research Ethics Committee of the USP Biosciences Institute. The parameters of the model to be developed in phase 2 will be fitted to reproduce the results obtained in the data collection, considering the sleep-wake cycle (homeostatic process), the circadian cycle, sleep inertia and workload. This set of parameters and their respective standard deviations will be used to estimate fatigue and sleepiness levels in a challenging operational environment.Expected results: The expected results have a highly relevant impact on the development of an innovative national product that can make accurate predictions of fatigue and sleepiness considering real work routines in complex operational environments with high cognitive demand. The influence of workload will also be assessed through a simultaneous fit of all data. As a result, the product resulting from this project will have improved functionalities and characteristics in relation to software already established globally. Both because of its characteristic of quantitatively reproducing all data with the same set of parameters customized to each operational environment, and because it provides confidence intervals for its estimates. (AU)