Can computers measure the chronic disease burden using survey questionnaires? The Symptomatic Diagnosis Study

Abstract Background Measuring the burden of chronic conditions in the developing world is a critical global health challenge. Scarce resources and the lack of biometry tests for chronic conditions such as depression and arthritis limit current measurement. Computer-based diagnosis based on self-reported signs and symptoms (symptomatic diagnosis) is a promising method for more accurately measuring the chronic disease burden. Methods As part of the Population Health Metrics Research Consortium (PHMRC) study, we collected 1379 questionnaires in Mexico from individuals who had a chronic condition that had been diagnosed with gold-standard diagnostic criteria or individuals who did not have any of the ten target conditions. Our primary analytical goal was to develop an algorithm to accurately diagnose chronic conditions. To this end, we tested methods previously developed for verbal autopsy. We analysed the performance of each method and compared performance to existing epidemiological measurement techniques. Findings The top-performing method is capable of achieving 68% concordance with gold-standard diagnosis. Concordance ranged from approximately 90% for depression, angina pectoris, and cirrhosis, to 40% for osteoarthritis and vision loss. The prevalence fraction of each condition could be measured with less than 3% absolute error. These findings roughly parallel validated verbal autopsy methods at an expectedly higher performance level. Interpretation Symptomatic diagnosis outperforms current techniques and is a viable method for measuring the burden of chronic diseases in areas with low health information infrastructure. It is a critical global health challenge to better characterise the epidemiology of chronic conditions in these areas, and is a powerful, unique solution capable of collecting an array of prevalence data in a single survey. This technology can provide myriad benefits to the field of epidemiology, including higher-resolution prevalence data, flexible data collection with rapid interpretability, and individual diagnosis for certain conditions. Funding The Population Health Metrics Research Consortium funded the data collection as part of a Gates Grand Challenges in Global Health initiative (GC-13).