Sebastien Tremblay

Sebastien Tremblay

Auto-generated profile, based on publicly available information

Sebastien Tremblay

Université Laval

Research Focus & Metrics

Academic Career Length

26Years (2001-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
21 December 2022 in JMIR Formative Research
Verified
Journal Article
Open access
Joel GagnonJoel Gagnon
Joel Gagnon
MKMichelle Khau
Michelle Khau
LLLéandre Lavoie-Hudon
Léandre Lavoie-Hudon
FVFrançois Vachon
François Vachon
VDVicky Drapeau
Vicky Drapeau
STSébastien Tremblay
Sébastien Tremblay
Background: Wearable devices collect physiological and behavioral data that have the potential to identify individuals at risk of declining mental health and well-being. Past research has mainly focused on assessing the accuracy and the agreement of heart rate (HR) measurement of wearables under different physical exercise conditions. However, the capacity of wearables to sense physiological changes, assessed by increasing HR, caused by a stressful event has not been thoroughly studied.Objective: This study followed 3 objectives: (1) to test the ability of a wearable device (Fitbit Versa 2) to sense an increase in HR upon induction of psychological stress in the laboratory; (2) to assess the accuracy of the wearable device to capture short-term HR variations caused by psychological stress compared to a gold-standard electrocardiogram (ECG) measure (Biopac); and (3) to quantify the degree of agreement between the wearable device and the gold-standard ECG measure across different experimental conditions.Methods: Participants underwent the Trier Social Stress Test protocol, which consists of an oral phase, an arithmetic stress phase, an anticipation phase, and 2 relaxation phases (at the beginning and the end). During the stress protocol, the participants wore a Fitbit Versa 2 and were also connected to a Biopac. A mixed-effect modeling approach was used (1) to assess the effect of experimental conditions on HR, (2) to estimate several metrics of accuracy, and (3) to assess the agreement: the Bland-Altman limits of agreement (LoA), the concordance correlation coefficient, the coverage probability, the total deviation index, and the coefficient of an individual agreement. Mean absolute error and mean absolute percent error were calculated as accuracy indices.Results: A total of 34 university students were recruited for this study (64% of participants were female with a mean age of 26.8 years, SD 8.3). Overall, the results showed significant HR variations across experimental phases. Post hoc tests revealed significant pairwise differences for all phases. Accuracy analyses revealed acceptable accuracy according to the analyzed metrics of accuracy for the Fitbit Versa 2 to capture the short-term variations in psychological stress levels. However, poor indices of agreement between the Fitbit Versa 2 and the Biopac were found.Conclusions: Overall, the results support the use of the Fitbit Versa 2 to capture short-term stress variations. The Fitbit device showed acceptable levels of accuracy but poor agreement with an ECG gold standard. Greater inaccuracy and smaller agreement were found for stressful experimental conditions that induced a higher HR. Fitbit devices can be used in research to measure HR variations caused by stress, although they cannot replace an ECG instrument when precision is of utmost importance.

9 August 2016 in Journal of Cognitive Engineering and Decision Making
Verified
Journal Article
Closed access
DLDaniel Lafond
Daniel Lafond
BRBenoît Roberge-Vallières
Benoît Roberge-Vallières
FVFrançois Vachon
François Vachon
STSébastien Tremblay
Sébastien Tremblay
Policy capturing is a judgment analysis method that typically uses linear statistical modeling to estimate expert judgments. A variant to this technique is to capture decision policies using data-mining algorithms designed to handle nonlinear decision rules, missing attributes, and noisy data. In the current study, we tested the effectiveness of a decision-tree induction algorithm and an instance-based classification method for policy capturing in comparison to the standard linear approach. Decision trees are relevant in naturalistic decision-making contexts since they can be used to represent “fast-and-frugal” judgment heuristics, which are well suited to describe human cognition under time pressure. We examined human classification behavior using a simulated naval air defense task in order to empirically compare the C4.5 decision-tree algorithm, the k-nearest neighbors algorithm, and linear regression on their ability to capture individual decision policies. Results show that C4.5 outperformed the other methods in terms of goodness of fit and cross-validation accuracy. Decision-tree models of individuals’ judgment policies actually classified contacts more accurately than their human counterparts, resulting in a threefold reduction in error rates. We conclude that a decision-tree induction algorithm can yield useful models for training and decision support applications, and we discuss the application of judgmental bootstrapping in real time in dynamic environments.

1 September 2016
Verified
Conference Paper
Closed access
ACAndrea Clerico
Andrea Clerico
CCCindy Chamberland
Cindy Chamberland
MPMark Parent
Mark Parent
PMPierre-Emmanuel Michon
Pierre-Emmanuel Michon
STSebastien Tremblay
Sebastien Tremblay
PJPhilip L Jackson
Philip L Jackson
The key to the development of adaptive gameplay is the capability to monitor and predict in real time the players experience (or, herein, fun factor). To achieve this goal, we rely on biometrics and machine learning algorithms to capture a physiological signature that reflects the player's affective state during the game. In this paper, we report research and development effort into the real time monitoring of the player's level of fun during a commercially available video game session using physiological signals. The use of a triple-classifier system allows the transformation of players' physiological responses and their fluctuation into a single yet multifaceted measure of fun, using a non-linear gameplay. Our results suggest that cardiac and respiratory activities provide the best predictive power. Moreover, the level of performance reached when classifying the level of fun (70% accuracy) shows that the use of machine learning approaches with physiological measures can contribute to predicting players experience in an objective manner.

19 February 2015 in JMIR Research Protocols
Verified
Journal Article
Open access
PAPatrick M Archambault
Patrick M Archambault
ATAlexis F Turgeon
Alexis F Turgeon
HWHolly O Witteman
Holly O Witteman
FLFrançois Lauzier
François Lauzier
LMLynne Moore
Lynne Moore
CGCanadian Critical Care Trials Group
Canadian Critical Care Trials Group
Background: Trauma is the most common cause of mortality among people between the ages of 1 and 45 years, costing Canadians 19.8 billion dollars a year (2004 data), yet half of all patients with major traumatic injuries do not receive evidence-based care, and significant regional variation in the quality of care across Canada exists. Accordingly, our goal is to lead a research project in which stakeholders themselves will adapt evidence-based trauma care knowledge tools to their own varied institutional contexts and cultures. We will do this by developing and assessing the combined impact of WikiTrauma, a free collaborative database of clinical decision support tools, and Wiki101, a training course teaching participants how to use WikiTrauma. WikiTrauma has the potential to ensure that all stakeholders (eg, patients, clinicians, and decision makers) can all contribute to, and benefit from, evidence-based clinical knowledge about trauma care that is tailored to their own needs and clinical setting. Objective: Our main objective will be to study the combined effect of WikiTrauma and Wiki101 on the quality of care in four trauma centers in Quebec. Methods: First, we will pilot-test the wiki with potential users to create a version ready to test in practice. A rapid, iterative prototyping process with 15 health professionals from nonparticipating centers will allow us to identify and resolve usability issues prior to finalizing the definitive version for the interrupted time series. Second, we will conduct an interrupted time series to measure the impact of our combined intervention on the quality of care in four trauma centers that will be selected-one level I, one level II, and two level III centers. Participants will be health care professionals working in the selected trauma centers. Also, five patient representatives will be recruited to participate in the creation of knowledge tools destined for their use (eg, handouts). All participants will be invited to complete the Wiki101 training and then use, and contribute to, WikiTrauma for 12 months. The primary outcome will be the change over time of a validated, composite, performance indicator score based on 15 process performance indicators found in the Quebec Trauma Registry. Results: This project was funded in November 2014 by the Canadian Medical Protective Association. We expect to start this trial in early 2015 and preliminary results should be available in June 2016. Two trauma centers have already agreed to participate and two more will be recruited in the next months. Conclusions: We expect that this study will add important and unique evidence about the effectiveness, safety, and cost savings of using collaborative platforms to adapt knowledge implementation tools across jurisdictions.

1 December 2019 in Journal of Experimental Psychology: Applied
Verified
Journal Article
Closed access
KLKatherine Labonté
Katherine Labonté
STSébastien Tremblay
Sébastien Tremblay
FVFrançois Vachon
François Vachon
In complex dynamic work environments, the consequences of task interruptions on performance can put public safety at risk. If not designed carefully, current tools aiming to facilitate interruption recovery can instead hamper performance because of information overload. Although a simpler solution-the forewarning of an imminent interruption-has proven effective in static contexts, existing theories of task interruption do not clearly predict its impact on the resumption of dynamically evolving tasks. The current study examined the effects of a preinterruption warning in dynamic settings to develop a better understanding of task resumption and supplement current theoretical accounts. In a simulation of above-water warfare, scenarios were either uninterrupted, unexpectedly interrupted, or interrupted following an auditory warning. Behavioral, oculomotor, and pupillometric data regarding decision making, information processing, and cognitive load were computed before, during, and after each interruption (or the corresponding moment). Interruption warnings triggered a cognitively demanding preinterruption preparation that, in turn, speeded up postinterruption information processing and decision making and lowered cognitive load when resuming the interrupted task. These findings help to complement current theories of interruptions while showing that preinterruption warnings represent a promising way to support interruption recovery in complex dynamic situations. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

1 January 2003 in Journal of Experimental Psychology: Applied
Verified
Journal Article
Closed access
SBSimon Banbury
Simon Banbury
LFLiz Fricker
Liz Fricker
STSébastien Tremblay
Sébastien Tremblay
LELucy Emery
Lucy Emery
Three experiments were conducted to examine the effects of extraneous speech warnings (i.e., low-priority warnings initiated during high-priority tasks) on cognitive performance and whether organizing the auditory warnings into streams can attenuate any disruption. Experiment 1 demonstrated that a variety of speech warnings can be separated into perceptually distinct streams by allocating them to discrete spatial locations. Experiment 2 showed that increasing the rate of presentation of the warnings to promote streaming decreased clarity ratings but increased perceived urgency ratings. Experiment 3 demonstrated that the disruption to serial memory for navigational information by extraneous speech warnings could be attenuated by streaming. Results are interpreted in light of previous research, and practical implications for auditory warning design are discussed.