Suicide is a major public health crisis in America, resulting in one death every 11 minutes, on average.
In 2020, suicide was among the top 9 leading causes of death for people between 10 and 64 years of age. Each year, suicide and nonfatal self-harm cost the nation nearly $490 billion in medical costs, work-loss costs, the value of statistical life, and quality-of-life costs, according to statistics cited by the U.S. Centers for Disease Control and Prevention.
For America’s veterans, the problem of suicide is even more grave. For example, in each year from 2001 through 2020, age- and sex-adjusted suicide rates of veterans exceeded those of non-veteran U.S. adults, according to the U.S. Department of Veterans Affairs (VA). Among veterans between the ages of 18 to 44, suicide was the second-leading cause of death.
To counter this serious issue, the VA in 2018 launched its “National Strategy for Preventing Veteran Suicide,” a roadmap for its plan to address the tragedy of suicide among veterans. A key piece of this wide-ranging initiative involves using synthetic data to build models that identify risk, tailor recommended treatments, and engage care to at-risk veterans as early as possible.
Why synthetic data?
Because veteran suicide is a complex, multifactored issue, VA leadership knew that achieving suicide-prevention goals would require an innovative, data-driven approach. This led to the VA employing synthetic data as one means of better understanding and reducing veteran suicide.
Synthetic data represents a form of data anonymization. Think of it as a way of taking private patient information and enabling researchers and other users to access the information contained within the data without compromising patient privacy. This process allows for greater utility to users of private patient data by creating a statistical model and populating it with new data points, novel patients, and synthetic patients.
The end result of that model is a data set that has the same statistical properties as the original data but doesn’t contain any of the original patients and therefore does not compromise their privacy. However, despite the absence of identifying patient information, synthetic data delivers full utility of the data because researchers can freely explore the information.
Synthetic data offers the potential to mimic the characteristics of a real dataset, without sensitive patient information, making it a good option for analyzing large but sensitive samples of real individual-level patient data. Synthetic data differs from de-identified data in that it is built from scratch, as opposed to being based on individual patient records, which means synthetic data cannot be de-anonymized.
Predicting mental health distress
The VA has leveraged synthetic data based on the real needs of various veteran patient populations to deliver an accurate picture of overall needs, and trends across populations at high suicide risk, and to proactively identify veterans at increasing risk of a suicide attempt.
Specifically, the VA has used synthetic data in the following four areas to predict suicide risk and advance suicide prevention:
1. Predictive modeling: By clarifying and confirming the mental-health challenges veterans are confronting, predictive modeling helps the VA understand how and where to focus suicide-prevention efforts.
2. Quality improvement: The VA is using synthetic data to understand clinical variation – the use of different healthcare services and practices – to identify which options have yielded the best outcomes in terms of mental health improvement and suicide reduction.
3. Population health management: Through population health management that relies on synthetic data, the VA can identify veterans at high risk for suicide who have been undertreated for mental health concerns.
4. Optimizing diagnosis and treatment: With new data sets produced via synthetic data, the VA can pinpoint trends that signify undiagnosed mental health concerns that are likely to increase veterans’ risk of suicide or self-harm.
Like many healthcare organizations, the VA experienced challenges with fragmented patient data and stringent, yet essential, privacy regulations. To overcome those roadblocks, the VA turned to synthetic data to accelerate its capabilities toward identifying and predicting veteran suicide risk and optimizing treatment. As a result of the program’s early success, the VA is expanding its synthetic data initiatives to include new models to improve treatment for heart disease, sleep apnea, and Alzheimer’s disease.
About Josh Rubel
Josh Rubel leads MDClone’s commercial team with a focus on building relationships with public and private health systems, life science, and health plan organizations through direct sales, partners, and channel alliances. Prior to joining MDClone, Josh spent 20 years in both established and new venture healthcare IT organizations including commercial leadership roles at GE Healthcare, Optum, and Enli Health Intelligence.