What your Fitbit Says about You: De-anonymizing Users in Lifelogging Datasets

Kazlouski, Andrei; Marchioro, Thomas; Markatos, Evangelos P.

doi:10.5220/0011268600003283

“…step counts) is a good predictor of a person's gender, BMI, and age, which can thus be inferred, despite being hidden during training [60]. (3) Overall, diabetes patients have the largest bias gap compared to their non-diabetic counterparts, partially attributed to their highly biased training data to start with.…”

Section: Aggregation Biasmentioning

confidence: 99%

“…Digital biomarkers contain an uncanny amount of personal information. Even the coarser behavioral biomarkers acquired from consumer wearable devices (such as steps, burned calories, and covered distance), strongly correlate to a person's gender, height, and weight [60], while signals of finer granularity (such as accelerometer and heart rate measurements), can predict variables associated with an individual's physical health, fitness, and demographics [91]. Similarly, (1) What does bias mean for PI?…”

Section: Introductionmentioning

confidence: 99%

Uncovering Bias in Personal Informatics

Yfantidou¹,

Sermpezis²,

Vakali³

et al. 2023

Preprint

0

View full text Add to dashboard Cite

Personal informatics (PI) systems, powered by smartphones and wearables, enable people to lead healthier lifestyles by providing meaningful and actionable insights that break down barriers between users and their health information. Today, such systems are used by billions of users for monitoring not only physical activity and sleep but also vital signs and women's and heart health, among others. Despite their widespread usage, the processing of sensitive PI data may suffer from biases, which may entail practical and ethical implications. In this work, we present the first comprehensive empirical and analytical study of bias in PI systems, including biases in raw data and in the entire machine learning life cycle. We use the most detailed framework to date for exploring the different sources of bias and find that biases exist both in the data generation and the model learning and implementation streams. According to our results, the most affected minority groups are users with health issues, such as diabetes, joint issues, and hypertension, and female users, whose data biases are propagated or even amplified by learning models, while intersectional biases can also be observed. CCS Concepts: • Human-centered computing → Ubiquitous and mobile computing; • Applied computing → Consumer health; • Computing methodologies → Artificial intelligence; • Social and professional topics → Codes of ethics.

show abstract

Practical Crowdsourcing of Wearable IoT Data with Local Differential Privacy

Marchioro

¹

,

Kazlouski

²

,

Markatos

³

2023

Proceedings of the 8th ACM/IEEE Conference on Internet of Things Design and Implementation

3

0

View full text Add to dashboard Cite

Uncovering Bias in Personal Informatics

Yfantidou¹,

Sermpezis²,

Vakali³

et al. 2023

Proc. ACM Interact. Mob. Wearable Ubiquitous Technol.

2

0

View full text Add to dashboard Cite

Personal informatics (PI) systems, powered by smartphones and wearables, enable people to lead healthier lifestyles by providing meaningful and actionable insights that break down barriers between users and their health information. Today, such systems are used by billions of users for monitoring not only physical activity and sleep but also vital signs and women's and heart health, among others. Despite their widespread usage, the processing of sensitive PI data may suffer from biases, which may entail practical and ethical implications. In this work, we present the first comprehensive empirical and analytical study of bias in PI systems, including biases in raw data and in the entire machine learning life cycle. We use the most detailed framework to date for exploring the different sources of bias and find that biases exist both in the data generation and the model learning and implementation streams. According to our results, the most affected minority groups are users with health issues, such as diabetes, joint issues, and hypertension, and female users, whose data biases are propagated or even amplified by learning models, while intersectional biases can also be observed.

show abstract

What your Fitbit Says about You: De-anonymizing Users in Lifelogging Datasets

Cited by 3 publications

References 0 publications

Uncovering Bias in Personal Informatics

Uncovering Bias in Personal Informatics

Practical Crowdsourcing of Wearable IoT Data with Local Differential Privacy

Uncovering Bias in Personal Informatics

Contact Info

Product

Resources

About