Obfuscation Methods for Wearable Electronic Devices

(EAT Trial)

Overseen By Nabil Alshurafa, PhD

Age: 18+

Sex: Any

Trial Phase: Academic

Sponsor: Northwestern University

Disqualifiers: Pregnant, Weight change, Genetic obesity, others

No Placebo Group All trial participants will receive the active study treatment (no placebo)

Compensation: Varies

Number of Visits: 3

72 Participants Needed

What You Need to Know Before You Apply

What is the purpose of this trial?

This trial tests a new wearable device designed to capture images of daily activities while preserving privacy. Worn around the neck, the device uses filters such as blur (Blur Obfuscation), cartoon (Cartoon Obfuscation), and edge effects (Edge Obfuscation) to obscure faces and objects in the images. Participants will use the device for four weeks, logging their food and drink intake and having phone check-ins with a dietitian. This study suits residents of the Chicago area who speak English and have a smartphone and computer. Participants will assist researchers in evaluating the effectiveness of the privacy features and provide feedback on the device. As an unphased trial, this study offers participants the chance to contribute to innovative research on privacy-preserving technology.

Do I need to stop taking my current medications for this trial?

The trial protocol does not specify whether you need to stop taking your current medications, so it's best to check with the study team for guidance.

What prior data suggests that this wearable device is safe for monitoring human behaviors?

Research shows that wearable devices, like those used in this study, are generally safe, though they can raise privacy concerns. Treatments such as Blur, Cartoon, and Edge Obfuscation focus on protecting privacy.

Cartoon Obfuscation reduces privacy worries, especially for nearby individuals. Blur and Edge Obfuscation help conceal identities in images, providing extra privacy protection. No safety issues have been reported with these visual filters; concerns focus more on data usage and security.

In this trial, the device's primary function is to take pictures silently. Various obfuscation methods alter these images to protect privacy. Other studies have not reported harmful effects from wearing such devices.

This trial is in the "Not Applicable" phase, indicating that safety is not a major concern. This phase typically involves non-invasive technology or trials that don't pose significant health risks. While privacy remains important, previous studies support the physical safety of wearing the device. ^{1 2 3 4 5}

Why are researchers excited about this trial?

Researchers are excited about this trial because it explores innovative methods to enhance privacy in wearable electronic devices. The trial investigates unique obfuscation techniques like Blur Obfuscation, Cartoon Obfuscation, and Edge Obfuscation, which are distinct from current approaches that simply restrict or encrypt data. Blur Obfuscation adds a layer of distortion to the data, while Cartoon Obfuscation transforms real images into cartoon-like representations, making it harder to extract sensitive information. Edge Obfuscation focuses on altering the outlines or edges of images, offering another layer of privacy protection. These methods aim to safeguard user privacy without compromising the functionality of wearable devices, which could be a game-changer in how personal data is protected in the digital age.

What evidence suggests that these obfuscation methods are effective for privacy preservation?

This trial will compare different obfuscation methods for wearable electronic devices. Research has shown that transforming images into cartoon-like visuals, as participants in the Cartoon Obfuscation arm will experience, can reduce privacy concerns, especially for those near the device user. This method makes facial recognition more difficult while still collecting activity data. Participants in the Edge Obfuscation arm will experience outlining objects with simple lines, which studies suggest helps protect privacy by obscuring specific details. Although blur obfuscation, another method tested in this trial, is commonly used, it often proves less effective at fully protecting privacy compared to other methods. These techniques aim to balance privacy protection with the need to gather useful activity data. ^{6 7 8 9 10}

Are You a Good Fit for This Trial?

This trial is for English-speaking individuals with a BMI of at least 18.5, living in the Chicago area, who have a computer, smartphone, and valid phone number. It's designed to assess a wearable device that monitors eating behaviors.

Inclusion Criteria

BMI greater than or equal to 18.5

Chicago metropolitan area resident

I have a working phone number.

Exclusion Criteria

Member of household enrolled in the study

Pregnant or breastfeeding individuals

I or someone in my family has a genetic obesity syndrome.

Timeline for a Trial Participant

Screening

Participants are screened for eligibility to participate in the trial

1 week

1 visit (in-person)

Device Training and Initial Recording

Participants visit the lab for consent, device training, and recording of several activities using the device

1 week

1 visit (in-person)

Active Weeks

Participants wear the device, keep a log of all food and drink items consumed, and participate in unscheduled phone calls with a dietitian

4 weeks

Unscheduled phone calls

Washout Weeks

Participants do not need to complete any study tasks during washout weeks

3 weeks

Feedback and Device Return

Participants return the device and provide feedback on the design and privacy-preserving features

1 week

1 visit (in-person)

What Are the Treatments Tested in This Trial?

Interventions

• Blur Obfuscation
• Cartoon Obfuscation
• Edge Obfuscation

Trial Overview

The study tests a neck-worn device capturing images without audio to monitor eating habits. Participants will use it during normal activities for four weeks with breaks in between. The device uses privacy filters (blur, edge, cartoon) on recorded images.

How Is the Trial Designed?

Group I: No Obfuscation (Raw) Experimental Treatment 1 Intervention

Group II: Edge Obfuscation Experimental Treatment 1 Intervention

Group III: Cartoon Obfuscation Experimental Treatment 1 Intervention

Group IV: Blur Obfuscation Experimental Treatment 1 Intervention

Who Is Running the Clinical Trial?

Northwestern University

Lead Sponsor

Trials

1,674

Recruited

989,000+

National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)

Collaborator

Trials

2,513

Recruited

4,366,000+

Published Research Related to This Trial

The study presents a novel method for teaching anatomical variations using 3D video animations, which enhance understanding of spatial relationships compared to traditional photographs or illustrations.

An aberrant right subclavian artery (ARSA) was identified during a dissection, and a 3D model of this variation was created, allowing students to interactively explore and understand its anatomical context on mobile devices.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Presentation of Anatomical Variations Using the Aurasma Mobile App. Hong, T., Bézard, G., Lozanoff, BK., et al.[2018]

The digital mask effectively removes identifiable features from facial images while preserving important disease-related characteristics necessary for diagnosis, enhancing patient privacy.

This innovation not only prevents recognition by both human researchers and facial-recognition algorithms but also increases patients' willingness to share their medical information, promoting better healthcare communication.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Anonymizing facial images to improve patient privacy. [2022]

A new display method using mosaic blur was developed to protect patient privacy in electronic medical records, allowing temporary concealment of sensitive information while still enabling necessary information sharing among medical personnel.

Interviews with medical staff revealed that while the mosaic blur increased awareness of patient confidentiality, it also posed challenges, such as the inconvenience of accessing concealed information and the potential risk of not having enough information for safe medical care.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Examination of an Electronic Patient Record Display Method to Protect Patient Information Privacy. Niimi, Y., Ota, K.[2017]

The article outlines the specific fabrication procedures for creating a one-way valved speech bulb obturator, which is a device used to improve speech in individuals with certain speech disorders.

This obturator is designed to enhance airflow control during speech, potentially leading to better communication outcomes for patients with velopharyngeal insufficiency.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

One-way valved speech bulb obturator using a tracheoesophageal prosthesis. Sutton, AJ., Hansen, NA., Myers, KV.[2013]

A new technique for creating custom face masks has been developed specifically for studying speech disorders related to craniofacial defects, using a method similar to dental lab procedures.

The masks are made from heat-cured acrylic resin and a resilient liner, ensuring they are both durable and comfortable for individuals during aeromechanical studies.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Custom-fabricated masks for aeromechanical measures. Kastner, CU., Putnam, AH., Shelton, RL.[2013]

The adapted double-flask technique is specifically designed for creating hollow maxillary obturators for patients with an atrophic maxillary alveolar ridge and a larger than normal interocclusal distance.

This technique may not be suitable for patients with a smaller interocclusal distance, as it limits the ability to fabricate a hollow denture effectively.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

The hollow denture: an alternative treatment for atrophic maxillae. Fattore, LD., Fine, L., Edmonds, DC.[2019]

DataSifter is a new statistical obfuscation method designed to de-identify sensitive high-dimensional data, such as clinical data from electronic health records, without sacrificing data utility.

Simulation results indicate that DataSifter effectively balances the risk of re-identification with the preservation of useful information, demonstrating its potential through practical applications on large datasets, like those related to autism.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

HDDA: DataSifter: statistical obfuscation of electronic health records and other sensitive datasets. Marino, S., Zhou, N., Zhao, Y., et al.[2023]

The new workflow integrates 3D digital technologies to create single-piece hollow obturator prostheses, eliminating the issues associated with conventional methods that often involve joints.

This innovative approach utilizes an optical 3D scanner and a laminating molding device, enhancing the efficiency and functionality of prosthesis fabrication.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

A workflow for fabricating a hollow obturator by using 3D digital technologies. Koyama, S., Kato, H., Harata, T., et al.[2020]

Activity-oriented partial obfuscation effectively maintains the ability to visually confirm hand-related activities while significantly enhancing bystander privacy, as shown in a study with 367 participants.

This new obfuscation method allows for the recording of hand activities without compromising the privacy of bystanders, making it a promising approach for researchers studying real-world interactions.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

To Mask or Not to Mask? Balancing Privacy with Visual Confirmation Utility in Activity-Oriented Wearable Cameras. Alharbi, R., Tolba, M., Petito, LC., et al.[2020]

The Hiding In Plain Sight (HIPS) method can effectively conceal about 90% of residual identifiers in clinical text, which is crucial for improving the safety of de-identification processes.

If validated further, HIPS could enhance existing de-identification systems, potentially increasing their effectiveness to over 99% without needing to improve their recall rates, making it a promising low-cost solution for protecting patient privacy.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Hiding in plain sight: use of realistic surrogates to reduce exposure of protected health information in clinical text. Carrell, D., Malin, B., Aberdeen, J., et al.[2021]

Citations

1.

withpower.com

withpower.com/trial/phase-wearable-electronic-device-2024-043a4

Obfuscation Methods for Wearable Electronic Devices

What data supports the effectiveness of the treatment Blur Obfuscation, Cartoon Obfuscation, Edge Obfuscation for wearable electronic devices? Research on ...

2.

pmc.ncbi.nlm.nih.gov

pmc.ncbi.nlm.nih.gov/articles/PMC9704364/

Impacts of Image Obfuscation on Fine-grained Activity ...

In this paper, we tested a mask obfuscation method's effect on hand-related human activity recognition accuracy. We show on average a 2% reduction in accuracy ...

3.

openaccess.thecvf.com

openaccess.thecvf.com/content_cvpr_2017_workshops/w16/papers/Caine_Blur_vs._Block_CVPR_2017_paper.pdf

Blur vs. Block: Investigating the Effectiveness of Privacy- ...

Results suggest that blurring (the most commonly used obfuscation) is much less effective at preserving privacy than blocking.

4.

par.nsf.gov

par.nsf.gov/servlets/purl/10373804

Impacts of Image Obfuscation on Fine-grained Activity ...

Our findings show that it is possible to protect PUI at smaller image utility costs (loss of accuracy). Index Terms—Human Activity Recognition, Wearable Camera,.

5.

researchgate.net

researchgate.net/publication/321637336_Effectiveness_and_Users'_Experience_of_Obfuscation_as_a_Privacy-Enhancing_Technology_for_Sharing_Photos

(PDF) Effectiveness and Users' Experience of Obfuscation ...

Results indicate the two most common obfuscations, blurring and pixelating, are ineffective. On the other hand, inpainting, which removes an ...

6.

mdpi.com

mdpi.com/2410-387X/8/2/26

“Whispers from the Wrist”: Wearable Health Monitoring ...

This manuscript explores the loopholes in current regulations and suggests a framework that categorizes wearable data and addresses challenges in data transfer.

7.

digitalhealthnews.com

digitalhealthnews.com/privacy-concerns-with-wearable-health-devices-what-you-need-to-know

Privacy Concerns with Wearable Health Devices

Despite their promise, wearable devices collect vast amounts of sensitive health data, making privacy a critical concern.

8.

aoshearman.com

aoshearman.com/en/insights/life-sciences-and-healthcare-insights/medical-wearables-under-the-microscope-us-regulatory-data-privacy-and-cybersecurity-perspectives

Medical wearables under the microscope: U.S. regulatory, ...

Falsified or corrupted data streamed from a jailbroken device could induce erroneous clinical interventions or cause a device to stop ...

9.

pmc.ncbi.nlm.nih.gov

pmc.ncbi.nlm.nih.gov/articles/PMC11879279/

HabitSense: A Privacy-Aware, AI-Enhanced Multimodal ...

Wearable cameras provide an objective method to visually confirm and automate the detection of health-risk behaviors such as smoking and overeating, ...

10.

cureus.com

cureus.com/articles/332968-what-clinicians-should-tell-patients-about-wearable-devices-and-data-privacy-a-narrative-review

What Clinicians Should Tell Patients About Wearable ...

The result is that people using wearables may be unaware of how their data may be used or the risks associated with data breaches. Many ...