Back

Does de-identification of data from wearable Biometric Monitoring Technologies give us a false sense of security? A systematic review

Chikwetu, L.; Miao, Y.; Woldetensae, M.; Bell, D.; Dunn, J.; Goldenholz, D. M.

2022-10-06 health informatics
10.1101/2022.10.04.22280658 medRxiv
Show abstract

It remains unknown whether de-identifying wearable biometric monitoring data is sufficient to protect the privacy of individuals in the dataset. This systematic review seeks to shed light on this. We searched Web of Science, IEEE Xplore Digital Library, PubMed, Scopus, and the ACM Digital Library on December 6, 2021 (PROSPERO CRD42022312922). We also performed manual searches in journals of interest until April 12, 2022. Though our search strategy had no language restrictions, all retrieved studies were in English. We included studies demonstrating re-identification, identification, or authentication using data from wearables. Our search returned 17,625 studies, and 72 studies met our inclusion criteria. Our findings demonstrate that substantial re-identification risk exists in data from sensors generally not thought to generate identifiable information, such as the electrocardiogram and electromyogram. In many cases, only a small amount of data (1-300 seconds of recording) is sufficient for re-identification.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

1
PLOS Digital Health
● 106 papers in training set
Top 0.5%
11.9%
Similar papers in this journal
2
npj Digital Medicine
○ 118 papers in training set
Top 0.6%
9.7%
Similar papers in this journal
3
Scientific Reports
○ 3612 papers in training set
Top 5%
9.6%
Similar papers in this journal
4
Sensors
○ 43 papers in training set
Top 0.1%
7.8%
Similar papers in this journal
5
BMJ Open
● 601 papers in training set
Top 3%
6.7%
Similar papers in this journal
6
DIGITAL HEALTH
○ 17 papers in training set
Top 0.1%
5.5%
50% of probability mass above
7
Journal of Medical Internet Research
◐ 87 papers in training set
Top 0.5%
4.8%
8
PLOS ONE
● 5266 papers in training set
Top 33%
4.0%
9
JAMIA Open
◐ 42 papers in training set
Top 0.5%
3.5%
10
Journal of the American Medical Informatics Association
◐ 71 papers in training set
Top 1.0%
3.2%
11
JMIR mHealth and uHealth
◐ 11 papers in training set
Top 0.1%
3.2%
12
Frontiers in Digital Health
○ 24 papers in training set
Top 0.3%
2.8%
13
Advanced Biology
○ 29 papers in training set
Top 0.2%
1.9%
14
JMIR Research Protocols
◐ 21 papers in training set
Top 0.8%
1.1%
15
IEEE Journal of Biomedical and Health Informatics
● 37 papers in training set
Top 0.9%
1.1%
16
Translational Vision Science & Technology
● 39 papers in training set
Top 0.5%
1.0%
17
IEEE Transactions on Neural Systems and Rehabilitation Engineering
● 49 papers in training set
Top 0.9%
0.8%
18
Communications Medicine
○ 113 papers in training set
Top 4%
0.8%
19
eClinicalMedicine
○ 77 papers in training set
Top 2%
0.8%
20
JAMA Network Open
● 130 papers in training set
Top 4%
0.8%
21
Frontiers in Neuroscience
○ 256 papers in training set
Top 6%
0.8%
22
Advanced Functional Materials
○ 46 papers in training set
Top 1%
0.6%
23
Gait & Posture
○ 24 papers in training set
Top 0.4%
0.6%
24
The Lancet Digital Health
○ 25 papers in training set
Top 1%
0.6%
25
BMJ
● 51 papers in training set
Top 1%
0.6%
26
Advanced Materials
○ 56 papers in training set
Top 1%
0.6%
27
Journal of NeuroEngineering and Rehabilitation
○ 36 papers in training set
Top 0.7%
0.6%

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.