[Skip to Navigation]
Views 359
Citations 0
Comment & Response
February 12, 2020

Application of Basic Epidemiologic Principles and Electronic Health Records in a Deep Learning Prediction Model

Author Affiliations
  • 1Department of Dermatology, Seoul National University College of Medicine, Seoul, South Korea
  • 2Interdisciplinary Program for Biomedical Engineering, Seoul National University, Seoul, South Korea
JAMA Dermatol. 2020;156(4):473-474. doi:10.1001/jamadermatol.2019.4922

To the Editor We found the study by Wang et al1 to be an interesting read. It raises important issues regarding the data-driven approach in the field of dermatology, and some obstacles should be discussed before applying it clinically.

Technically, the imbalances between study groups can cause overfitting problems. The patients in the study who had nonmelanoma skin cancer were much older than the control patients, which led to many more annual diagnostic code counts and medications per person. The patterns of diagnosis and medications in the 2 groups could definitely be tied to age difference. The k-fold cross-validation method can reduce but not completely remove overfitting, especially for small data sets.2 Performance results for each of the test and validation sets were not shown separately. Therefore, external validation should have been conducted to verify whether the model of Wang et al1 predicted skin cancer only in elderly patients or in patients in the general population.

Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words