JMIR Publications

ABSTRACT

Background:

As the need for sharing of genomic data grows, privacy issues and concerns, such as the ethics surrounding data sharing and disclosure of personal information, are raised.

Objective:

The main purpose of this study was to verify whether genomic data is sufficient to predict a patient's personal information.

Methods:

Genomic RNA expression data and matched patient personal information were collected from 9,538 patients in The Cancer Genome Atlas. Five personal information variables (age, gender, race, cancer type, and stage) were recorded for each patient. Four machine learning algorithms were used to confirm whether a patient's personal information could be accurately predicted from RNA expression data. The performance measurement of the prediction models was based on the accuracy and area under the receiver operating characteristic curve (AUC). We selected five cancer types (BRCA, KIRC, HNSC, LGG, and LUAD) with large samples sizes to verify whether predictive accuracy would differ between cancer types. We also validated the efficacy of our four machine learning models in analyzing normal samples from 593 cancer patients.

Results:

In most samples, personal information with high genetic relevance, such as gender and cancer type, could be predicted from RNA expression data alone. The prediction accuracies for gender and cancer type, which were the best models, were 0.93–0.99 and 0.78–0.94, respectively. Other aspects of personal information, such as age, race, and stage, were difficult to predict from RNA expression data, with accuracies ranging from 0.0026–0.29, 0.76–0.96, and 0.45–0.79, respectively. Among the tested machine learning methods, the highest predictive accuracy was obtained using the support vector machine (average accuracy: 0.77) and the lowest accuracy was obtained using the random forest method (average accuracy: 0.65). Gender and race were more accurately predicted compared to other variables in the samples from the five cancers. On average, the accuracy of predicting stage ranged between 0.71-0.67 and the age prediction accuracy ranged from 0.18-0.23 for the five cancers.

Conclusions:

Although genomic data can be used to obtain some personal information from a patient sample, it is difficult to obtain information sufficient to predict each patient without overlap in the RNA sequencing dataset. In this study, we present the possibility of sharing genomic datasets while still protecting sensitive personal information.