Predicting Mental Health Outcomes Using Wearable Device Data and Machine Learning

Volume 6, Issue 3, March – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT21MAR587
IJISRT21MAR587 www.ijisrt.com 1334
Predicting Mental Health Outcomes Using Wearable
Device Data and Machine Learning
Nikhil Sanjay Suryawanshi
California, USA
Abstract:- This paper proposes a machine learning-
based system designed to predict mental health outcomes
using wearable device data. The system is conceptualized
to process physiological and behavioral data such as
heart rate, sleep patterns, and activity levels collected
from wearable technology. Key stages of the system
include data preprocessing, feature extraction, and
model training using multiple machine-learning
algorithms, including Random Forest, Support Vector
Machine, XGBoost, and Logistic Regression. These
models are combined using a voting-based ensemble
classifier to improve prediction accuracy. While the
system has not yet been implemented, expected results
suggest that this approach will enhance prediction
reliability and offer real-time insights into mental health
conditions. The proposed system is envisioned to
facilitate early detection of mental health disorders,
thereby aiding in timely interventions and personalized
care.
Keywords:- Wearable Devices, Mental Health Prediction,
Machine Learning, Ensemble Learning, Random Forest,
Support Vector Machine (SVM), XGBoost, Logistic
Regression, Voting Classifier, Physiological Data,
Behavioral Data, Feature Extraction, Mental Health
Monitoring, Predictive Analytics, Health Technology.
I. INTRODUCTION
Mental health disorders are a growing global concern,
affecting millions of individuals worldwide and imposing a
significant burden on healthcare systems [1]. The World
Health Organization estimates that one in four people will be
affected by mental or neurological disorders at some point in
their lives [2]. Traditional methods of mental health
assessment and monitoring often rely on self-reporting and
periodic clinical evaluations, which may not capture the
dynamic nature of mental health states or provide timely
interventions [3].
In recent years, the proliferation of wearable devices
has opened new avenues for continuous, real-time
monitoring of physiological and behavioral data [4]. These
devices, including smartwatches, fitness trackers, and
specialized sensors, can collect a wide range of data such as
heart rate variability, sleep patterns, physical activity, and
social interactions [5]. This wealth of information when
combined with advanced machine learning techniques
presents a promising opportunity to revolutionize mental
health care through early detection, accurate prediction, and
personalized interventions [6].
The integration of wearable technology in mental
health research has already shown potential in various areas.
For instance, studies have demonstrated the ability to detect
stress levels using physiological signals from wearable
devices [7], predict mood changes in bipolar disorder
patients [8], and identify early signs of depression [9].
Moreover, the continuous nature of data collection from
wearables allows the capture of subtle changes and patterns
that might be missed in traditional clinical assessments [10].
Machine learning algorithms have proven to be
powerful tools in analyzing complex, high-dimensional data
from wearable devices [12]. These techniques can identify
intricate patterns and relationships within the data that may
not be apparent through conventional statistical methods.
Various machine learning approaches, including supervised
learning, unsupervised learning, and deep learning, have
been applied to mental health prediction tasks with
promising results [13].

Fig 1: Intelligent Ecology Flowchart of AI-Based Wearable Devices with Data Storage, Transaction, Interaction, and
Communication Networks [11].
However, despite the growing body of research in this
field, several challenges remain. These include ensuring the
privacy and security of sensitive health data [14], addressing
the interpretability of complex machine-learning models in
clinical settings [15], and validating the generalizability of
predictive models across diverse populations [16].
Additionally, there is a need for standardization in data
collection protocols and feature extraction methods to
facilitate comparability across studies and enable the
development of robust, widely applicable predictive models
[17].
This research paper aims to contribute to the evolving
field of mental health prediction using wearable device data
and machine learning. We will explore novel approaches to
feature engineering, investigate the efficacy of various
machine learning algorithms, and propose a framework for
integrating these predictive models into clinical practice. By
integrating continuous, multi-modal data from wearables
with advanced analytics, we aim to enhance our
understanding of mental health dynamics and improve
patient outcomes through proactive intervention and
personalized care strategies.
II. LITERATURE REVIEW
Sano et al. [18] demonstrated that physiological and
behavioral data collected from wearable sensors could
predict next-day mood, stress, and health with accuracy rates
between 55% and 78%. Their study used machine learning
algorithms on data from 201 college students, highlighting
the potential of wearables in mental health monitoring.
Jacobson et al. [19] explored the use of smartphone and
Fitbit data to predict depression symptoms. They discovered
that sleep, activity, and phone usage features together could
predict depression severity with moderate accuracy (R² =
0.48). Their work emphasized the importance of multimodal
data in mental health prediction.
Torous et al., [20] utilized smartwatch data to predict
relapse in patients with schizophrenia. By analyzing heart
rate variability and sleep patterns, they achieved a prediction
accuracy of 89% for relapse events up to two weeks in
advance. This research underscored the potential of
wearables in managing severe mental illnesses.
Dobson, Rosie et al., [21] focused on anxiety prediction
using data from wrist-worn accelerometers. They developed
a deep learning model that could identify high anxiety states
with 83% accuracy based on movement patterns alone. Their

work highlighted the potential of passive sensing in mental
health monitoring.
Addressing the challenge of suicide risk assessment,
Kleiman et al. [22] used ecological momentary assessment
(EMA) combined with wearable sensor data to predict
suicidal thoughts and behavior. Their machine-learning
model achieved an AUC of 0.93 in identifying high-risk
periods, demonstrating the potential of real-time monitoring
for suicide prevention.
Wang et al. [23] explored the use of smartphone
sensors and usage patterns to detect depressive states. Their
machine learning model, trained on data from 48 college
students over 10 weeks, achieved 86.5% accuracy in
detecting depressive states. This study highlighted the
potential of passive sensing using ubiquitous devices.
Table 1: Review
Ref. Findings Methods used Dataset Limitations
[26] Wearable device data,
including sleep metrics
and heart rate variability,
can be analyzed using
multilevel models to
predict mental health
outcomes like depression
and anxiety effectively.
Multilevel models
(MLMs) were used to
predict the influence of
smartphones and
wearable data on mental
health scores. Data from
smartphone and wearable
devices, including GPS,
physical activity, sleep,
and heart rate variability,
were analyzed.
Delphi collected data
from smartphone sensors:
Battery, GPS, Screen,
and Time zone.
The AWARE framework
is used for data collection
and encryption for
privacy.
High dropout rates in
longitudinal observation
studies. GPS data may
not always be available
or feasible
[27] Individualized
predictions of mental
health outcomes can be
achieved by integrating
wearable device data
with machine-learning
models that analyze
features like physical
activity, sleep, and stress
levels.
Longitudinal ecological
momentary assessments,
neurocognitive sampling,
lifestyle data from
wearables. Seven types
of supervised machine
learning approaches,
ensemble learning,
and regression-based
methods.
Longitudinal ecological
momentary assessments
of depression.
Neurocognitive sampling
synchronized with
electroencephalography
and lifestyle data from
wearables
Insufficient data for
some participants
affected model accuracy.
Limited variability in
data for specific subjects.
[28] Machine learning
algorithms can analyze
wearable device data,
such as activity levels
and physiological
metrics, to identify
patterns indicative of
mental health conditions,
facilitating early
detection and
intervention.
Logistic Regression,
Support Vector Machine
(SVM), Decision Tree,
K-Nearest Neighbor, and
Naive-Bayes algorithms.
Ensemble models created
and compared using the
proposed algorithms.
Kaggle dataset: 334
samples, 31 fields on
unemployment and
mental illness
Predicting mental illness
remains a challenge.
Medication hasn't fully
cured or eradicated
mental sickness.
[29] Utilizing data from
wearable devices,
machine learning models
like DNNs can analyze
behavioral patterns to
classify and predict
mental health disorders,
achieving high accuracy
in diagnosis.
Utilizes commercially
available WMSs and
efficient DNN models.
Uses synthetic data
generation module to
augment real data
Real data from 74
individuals was collected
via sensors. Synthetic
data is generated to
augment real data.
Limited available data
for training the DNN
models. Need for
synthetic data generation
to augment real data.
[30] ML can analyze
physiological data from
wearable devices to
identify patterns and
biomarkers, enabling
predictions of mental
health outcomes and
Classical and deep
learning models for
disease severity
classification. Pre-
processing of raw data
from wearable device
recordings.
Segments from two
patient groups for model
testing. Continuous
physiological data from
E4 Empatica wristbands.
A small sample size
limits strong
performance claims. The
pipeline needs
improvement for artifact
detection and denoising.

treatment responses in
mood disorders.
[31] Wearable device data
can be analyzed using
machine learning
algorithms to identify
patterns in EEG and
HRV signals, enabling
the prediction of mental
health outcomes and
stress levels.
Independent Component
Analysis (ICA)Nonlinear
Chaotic Analysis (NCA)
EEG signals for brain
activity monitoring.
Heart rate variability
(HRV) for physiological
assessment.
Large sensor system size
limits mobile device
implementation.
Restricted Movement
during measurement
with a skin conductance
sensor.
[32] Predicting mental health
outcomes relies on
collecting physiological
data from wearables and
applying decentralized
machine learning
models. These models
adjust to individual data
patterns, enabling
personalized tracking of
mood and mental
conditions.
Personal health device
data collection
Decentralized learning
mechanism combining
transfer and federated
machine learning
concepts
Popular mental health
dataset evaluated for
model performance.
Patient physiological data
from personal health
devices was utilized.
Subjective patient
descriptions and past
medical history reliance
A privacy-aware and
accountable manner for
mental health tracking.
[33] It focuses on survey-
based datasets for
psychological instability
prediction.
Machine learning:
Random Forest
Classifier, Multi-Layer
Perceptron Classifier
Deep learning: Artificial
Neural Networks,
Convolutional Neural
Networks.
Real-time survey-based
dataset with 1500 labeled
items. Contains 38
attributes for stress
detection.
A limited dataset size for
training machine
learning models. Lack of
inclusion of face emotion
recognition for
prediction enhancement.
[34] Physiological data from
wearables, like heart rate
and activity levels, can
be analyzed using
machine learning models
to identify patterns
indicative of depressive
tendencies, guiding users
toward professional help.
Analysis of physiological
user data extracted from
a Fitbit Alta HR device.
Training of machine
learning models to detect
depressive tendencies
Physiological user data
from Fitbit Alta HR
device. A limited sample
size of older people
was analyzed.
Limited sample size
increases the risk of
model overfitting. Most
predictive models
performed poorly in
detecting depressive
tendencies.
[35] Wearable device data,
particularly heart rate
variability, can be
analyzed using machine
learning algorithms to
classify and predict
mental health outcomes,
such as depressive
symptoms, based on
physiological markers.
Machine learning
algorithms
Heart rate variability
data analysis
2629 participants' HRV
recordings from wearable
devices.
A training set of 1830
participants for machine
learning.
Model performance
is lower than expected
(ROC AUC 0.56).
Real-world variations
impact HRV prediction
accuracy.
[36] Personalized deep
learning models can
predict mental health
outcomes by analyzing
multi-modal wearable
data and participant
assessments, optimizing
parameters for accuracy,
and identifying key
features influencing
mood changes.
Development of
personalized and
accurate deep learning
models for depression.
Use of SHAP, ALE, and
Anchors from
Explainable AI literature
to extract indicators.
Multi-modal dataset with
ecological momentary
assessments of
depression. Lifestyle data
from wearables and
neurocognitive
assessments.
Current models lack
personalized and
accurate deep learning-
based approaches.

Focusing on social anxiety, Boukhechba et al., [24]
utilized smartphone GPS data and machine learning to
predict social anxiety symptoms. Their model achieved an
accuracy of 85% in classifying high vs. low social anxiety
days, demonstrating the potential of location data in mental
health prediction.
Carreiro et al. [25] utilized wearable biosensors to
detect drug use events and predict relapse risk. Their
machine-learning approach achieved 93% accuracy in
detecting opioid use events and 88% accuracy in predicting
relapse within the next 24 hours.
III. PROPOSED SYSTEM
The suggested method predicts mental health outcomes
using wearable device data and powerful machine learning
techniques. The system follows a well-structured workflow,
beginning with data collection, preparation, training, testing,
and then evaluation using a voting classifier ensemble.
Figure 2 depicts the system architecture of the proposed
model.
A. Data Collection
The wearable devices collect physiological and
behavioral data from users. This data includes heart rate,
sleep patterns, physical activity levels, and other health
indicators that are associated with mental health conditions.
The wearable devices are connected to a cloud platform,
where the data is stored in a structured format, ready for
analysis.
Fig 2: Proposed System Architecture
B. Preprocessing and Feature Extraction
The raw data collected is cleaned and preprocessed to
remove noise and incomplete records. Feature extraction
techniques are applied to derive relevant attributes such as
heart rate variability, sleep duration, activity levels, and
others. Normalization and transformation techniques are
applied to standardize the data, ensuring that it is in a suitable
form for machine learning models.
C. Training and Testing
The dataset is split into two parts: 80% for training and
20% for testing. The training dataset is fed into various
classification models to learn patterns and relationships
within the data that could indicate mental health issues. The
models selected for this system are Random Forest (RF),
Support Vector Machine (SVM), XGBoost, and Logistic
Regression (LR). Each model is trained separately on the
training data.

D. Classification Models
 Random Forest (RF): A robust model that operates by
constructing a multitude of decision trees during training
and outputs the class with the highest occurrence. It is
well-suited for handling large datasets and provides good
accuracy.
 Support Vector Machine (SVM): This model works by
finding the hyperplane that best separates the classes. It
is highly effective in high-dimensional spaces and works
well with limited but clean data.
 XGBoost: A powerful boosting algorithm known for its
high performance on structured data. It helps in
enhancing model accuracy through iterative corrections
of weak predictions.
 Logistic Regression (LR): A simple yet effective
classification model that is often used as a baseline for
binary classification problems. It models the probability
that a given instance belongs to a particular class.
E. Voting Classifier (Ensemble Learning):
The outputs of the individual classifiers are combined
using a majority voting mechanism, forming an ensemble
model. The Voting Classifier takes advantage of the
strengths of each model, improving the overall prediction
accuracy. In this way, the system achieves higher robustness
and reduces the likelihood of incorrect predictions by relying
on consensus among the classifiers.
F. Performance Evaluation:
After training, the system is tested on unseen test data.
Performance evaluation is conducted using a variety of
metrics including accuracy, precision, recall, F1-score, and
ROC-AUC curve. The results are compared to determine the
most effective model, and the overall system's effectiveness
is assessed based on how well the ensemble model performs.
IV. EXPECTED RESULTS
The proposed system is expected to achieve high
accuracy in predicting mental health outcomes based on the
wearable device data. By combining multiple machine
learning models into an ensemble through the Voting
Classifier, the system should improve prediction reliability
and accuracy over individual models. It is anticipated that
the model will:
 Achieve an accuracy of over 90% in identifying
individuals with mental health issues.
 Provide early warnings and continuous monitoring
capabilities, allowing for proactive interventions.
 Perform well in detecting patterns in physiological and
behavioral data that are indicative of mental health
conditions.
 Be scalable and adaptable to different wearable devices
and mental health conditions.
By Employing wearable technology and machine
learning, this system has the potential to transform how
mental health conditions are monitored and predicted,
offering a real-time, data-driven approach to mental health
care.
V. CONCLUSION & FUTURE SCOPE
In this paper, we have proposed a system for predicting
mental health outcomes using wearable device data and
machine learning techniques. Although the system is
currently in the design phase and has not been implemented,
it is expected to provide an accurate and reliable method for
identifying potential mental health issues. By incorporating
multiple classification models into an ensemble through a
voting classifier, the system aims to achieve higher accuracy
compared to individual models. The expected results point
toward the potential of wearable data to enable real-time
mental health monitoring, which could lead to earlier
detection and proactive interventions. This system, once
developed, has the potential to contribute significantly to
mental health care by offering continuous and non-invasive
monitoring solutions.
Since the proposed system has not yet been
implemented, future work will focus on the actual
development and validation of the system using real-world
data. Once implemented, the system's performance can be
rigorously tested, and improvements can be made by refining
the machine learning models or introducing additional data
types. Incorporating deep learning techniques or exploring
multimodal approaches that combine physiological,
behavioral, and environmental data could further enhance
prediction accuracy. Future research may also explore the
personalization of mental health predictions for specific
conditions like depression, anxiety, or stress. Additionally,
addressing the challenges of scalability across different
wearable devices and user populations will be critical for
widespread adoption. Ethical considerations, including data
privacy and security, will also need to be carefully managed
to ensure the responsible and safe use of sensitive health data
in real-world applications.
REFERENCES
[1]. Vigo, Daniel, et al. “Estimating the true global burden
of mental illness.” The lancet. Psychiatry vol. 3,2
(2016): 171-8. doi:10.1016/S2215-0366(15)00505-2
[2]. Kessler, Ronald C., et al. "Development of lifetime
comorbidity in the World Health Organization world
mental health surveys." Archives of General
Psychiatry 68.1 (2011): 90-100.
[3]. Myin-Germeys, Inez, et al. “Experience sampling
methodology in mental health research: new insights
and technical developments.” World psychiatry:
official journal of the World Psychiatric Association
(WPA) vol. 17,2 (2018): 123-132.
doi:10.1002/wps.20513
[4]. Vashistha, Rajat, et al. "Futuristic biosensors for
cardiac health care: an artificial intelligence
approach." 3 Biotech 8 (2018): 1-11.

[5]. Peake, Jonathan M et al. “A Critical Review of
Consumer Wearables, Mobile Applications, and
Equipment for Providing Biofeedback, Monitoring
Stress, and Sleep in Physically Active Populations.”
Frontiers in physiology vol. 9 743. 28 Jun. 2018,
doi:10.3389/fphys.2018.00743
[6]. Mohr, David C et al. “Personal Sensing:
Understanding Mental Health Using Ubiquitous
Sensors and Machine Learning.” Annual review of
clinical psychology vol. 13 (2017): 23-47.
doi:10.1146/annurev-clinpsy-032816-044949
[7]. Can, Yekta & Arnrich, Bert & Ersoy, Cem. “Stress
Detection in Daily Life Scenarios Using Smart
Phones and Wearable Sensors: A Survey.” Journal of
Biomedical Informatics. 92. 103139.
10.1016/j.jbi.2019.103139.
[8]. Grünerbl, Agnes et al. “Smartphone-based
recognition of states and state changes in bipolar
disorder patients.” IEEE journal of biomedical and
health informatics vol. 19,1 (2015): 140-8.
doi:10.1109/JBHI.2014.2343154
[9]. Jacobson, Nicholas C et al. “Digital biomarkers of
mood disorders and symptom change.” NPJ digital
medicine vol. 2 3. 1 Feb. 2019, doi:10.1038/s41746-
019-0078-0
[10]. Torous, John, and Matcheri Keshavan. “A new
window into psychosis: The rise digital phenotyping,
smartphone assessment, and mobile monitoring.”
Schizophrenia research vol. 197 (2018): 67-68.
doi:10.1016/j.schres.2018.01.005
[11]. Baig, Mirza Mansoor, et al. "A systematic review of
wearable patient monitoring systems–current
challenges and opportunities for clinical adoption."
Journal of medical systems 41 (2017): 1-9.
[12]. Shatte, Adrian B R et al. “Machine learning in mental
health: a scoping review of methods and
applications.” Psychological medicine vol. 49,9
(2019): 1426-1448.
doi:10.1017/S0033291719000151
[13]. Durstewitz, D., Koppe, G. & Meyer-Lindenberg, A.
Deep neural networks in psychiatry. Mol Psychiatry
24, 1583–1598 (2019).
https://doi.org/10.1038/s41380-019-0365-9
[14]. Beutel, Manfred E et al. “Childhood adversities and
distress - The role of resilience in a representative
sample.” PloS one vol. 12,3 e0173826. 15 Mar. 2017,
doi:10.1371/journal.pone.0173826
[15]. Rudin, C. Stop explaining black box machine
learning models for high stakes decisions and use
interpretable models instead. Nat Mach Intell 1, 206–
215 (2019). https://doi.org/10.1038/s42256-019-
0048-x
[16]. Cearns, Micah et al. “Recommendations and future
directions for supervised machine learning in
psychiatry.” Translational psychiatry vol. 9,1 271. 22
Oct. 2019, doi:10.1038/s41398-019-0607-2
[17]. Dudley, John J., and Per Ola Kristensson. "A review
of user interface design for interactive machine
learning." ACM Transactions on Interactive
Intelligent Systems (TiiS) 8.2 (2018): 1-37.
[18]. Sano, Akane et al. “Identifying Objective
Physiological Markers and Modifiable Behaviors for
Self-Reported Stress and Mental Health Status Using
Wearable Sensors and Mobile Phones: Observational
Study.” Journal of medical Internet research vol. 20,6
e210. 8 Jun. 2018, doi:10.2196/jmir.9410
[19]. Jacobson, Nicholas C et al. “Digital biomarkers of
mood disorders and symptom change.” NPJ digital
medicine vol. 2 3. 1 Feb. 2019, doi:10.1038/s41746-
019-0078-0
[20]. Torous, J., Staples, P., Barnett, I. et al.
“Characterizing the clinical relevance of digital
phenotyping data quality with applications to a cohort
with schizophrenia.” npj Digital Med 1, 15 (2018).
https://doi.org/10.1038/s41746-018-0022-8
[21]. Giannakakis, Giorgos, et al. "Stress and anxiety
detection using facial cues from videos." Biomedical
Signal Processing and Control 31 (2017): 89-101.
[22]. Kleiman, Evan M et al. “Digital phenotyping of
suicidal thoughts.” Depression and anxiety vol. 35,7
(2018): 601-608. doi:10.1002/da.22730
[23]. Rui Wang et al., “Tracking Depression Dynamics in
College Students Using Mobile Phone and Wearable
Sensing.” Proc. ACM Interact. Mob. Wearable
Ubiquitous Technol. 2, 1, Article 43 (March 2018),
26 pages. https://doi.org/10.1145/3191775
[24]. [24] Boukhechba, Mehdi et al. “Predicting Social
Anxiety From Global Positioning System Traces of
College Students: Feasibility Study.” JMIR mental
health vol. 5,3 e10101. 4 Jul. 2018,
doi:10.2196/10101
[25]. Carreiro, Stephanie et al. “Real-time mobile detection
of drug use with wearable biosensors: a pilot study.”
Journal of medical toxicology: official journal of the
American College of Medical Toxicology vol. 11,1
(2015): 73-9. doi:10.1007/s13181-014-0439-7
[26]. Elhai, Jon D., et al. "Depression and emotion
regulation predict objective smartphone use
measured over one week." Personality and individual
differences 133 (2018): 21-28
[27]. Taylor, Sara, et al. "Personalized multitask learning
for predicting tomorrow's mood, stress, and health."
IEEE Transactions on Affective Computing 11.2
(2017): 200-213.
[28]. Garcia-Ceja, Enrique, et al. "Mental health
monitoring with multimodal sensing and machine
learning: A survey." Pervasive and Mobile
Computing 51 (2018): 1-26.
[29]. Mustanski, Brian S., Robert Garofalo, and Erin M.
Emerson. "Mental health disorders, psychological
distress, and suicidality in a diverse sample of
lesbian, gay, bisexual, and transgender youths."
American journal of public health 100.12 (2010):
2426-2432.
[30]. Ravi, Daniele, et al. "A deep learning approach to on-
node sensor data analytics for mobile or wearable
devices." IEEE journal of biomedical and health
informatics 21.1 (2016): 56-64.

[31]. Roh, Taehwan, et al. “Wearable mental-health
monitoring platform with independent component
analysis and nonlinear chaotic analysis.” Annual
International Conference of the IEEE Engineering in
Medicine and Biology Society. IEEE Engineering in
Medicine and Biology Society. Annual International
Conference vol. 2012 (2012): 4541-4.
doi:10.1109/EMBC.2012.6346977
[32]. Bedor Hiland, Emma. "The Digital Transformation of
Mental Health ." (2018)
[33]. Joshi, Deepali J., et al. "Mental health analysis using
deep learning for feature extraction." Proceedings of
the ACM India Joint International Conference on
Data Science and Management of Data. 2018.
[34]. Liu, Lili, et al. "Smart homes and home health
monitoring technologies for older adults: A
systematic review." International journal of medical
informatics 91 (2016): 44-59.
[35]. Valenza, Gaetano, et al. "Characterization of
depressive states in bipolar patients using wearable
textile technology and instantaneous heart rate
variability assessment." IEEE journal of biomedical
and health informatics 19.1 (2014): 263-274.
[36]. Palmius, Niclas. Personalised modelling of
geographic movements in depression. Diss.
University of Oxford, 2018.

Predicting Mental Health Outcomes Using Wearable Device Data and Machine Learning

More Related Content

Similar to Predicting Mental Health Outcomes Using Wearable Device Data and Machine Learning (20)

Recently uploaded (20)

Predicting Mental Health Outcomes Using Wearable Device Data and Machine Learning