Data fidelity refers to the accuracy with which information—“data”—qualifies and demonstrates the characteristics of the origin—“source.” The emergence of artificial intelligence in data monitoring has necessitated an elevated level of accuracy.
We are discussing the specific topic of data fidelity and accuracy because of recent events surrounding the Department of Government Efficiency (DOGE) audits of the Social Security Administration master file, which found a significant number of presumably “deceased beneficiaries” receiving Social Security benefits (or not?). The most egregious case involved a beneficiary aged 360 years—someone receiving Social Security benefits from before the birth of this republic in 1776. (It happens, folks, and it happened to me—more on that later.) Although the terms data fidelity and data accuracy are frequently used interchangeably, they represent distinct and additional aspects of data quality.
In 2012, the Office of the Inspector General for the Department of Health and Human Services (OIG-HHS) published a report, “Medicare Payments Made on Behalf of Deceased Beneficiaries in 2011.” The report outlined how there was a clerical error at a rate of 0.01 percent in Medicare providers billing for beneficiaries who were deceased (though the Medicare system, since 2005, blocks any payment). It also detailed problems with the Social Security Master File and the Social Security Master Death File, explaining how maintaining accurate data within appropriate parameters is difficult.
Data quality refers to how accurate, precise, and reliable collected information is to the extent that a health care professional can trust and use the information to diagnose patients, prescribe treatments, and create treatment plans. An example is high-fidelity electrocardiogram recordings. Similarly, information and data accuracy focus on the authenticity of data points, ensuring that measurements truly represent the actual parameters and information being monitored. A practical example is patient monitoring systems that capture a patient’s blood pressure every minute with high fidelity. However, if the device is improperly calibrated, these readings may lack accuracy despite their high fidelity. This distinction is very important, as it pertains to the reliability of data in decision-making.
The role of data accuracy and fidelity in contextualizing information can be demonstrated through the government’s use of data in health care prosecutions. In United States of America v. Muhamad Aly Rifai, I was accused of billing for services on deceased Medicare beneficiaries. The government experts were perplexed and speechless as they attempted to explain to the jury how 90 percent of the data that supposedly showed billing on deceased patients were for claims on beneficiaries who had died before I was born or during my years in kindergarten. (That 360-year-old American is not an oddity, folks.) The government’s case collapsed at trial when these data accuracy and fidelity issues were revealed to the jury. Similarly, the government bureaucracy became the target of ridicule by President Donald Trump and Elon Musk when the Social Security benefit payments for the 360-year-old human were published. This may have highlighted the importance of data fidelity and accuracy, as one aberrant data point cast doubt on the majority of the hard work done by federal employees at the Social Security Administration.
This brings us to the subject of data quality, which refers to the degree to which information is reliable and plays a role in maintaining data integrity. The quality of data is very important and depends on several factors. First, the information and data must be complete and contain the most accurate version of the information required. The information should be consistently validated to ensure that the data is formatted uniformly. In the case of DOGE’s reporting on deceased Social Security beneficiaries receiving benefits, the discrepancies may have resulted from unique incorrect entries that skewed the results in a disastrous way. This demonstrates how the relationship between data fidelity and data quality is particularly significant, as high-quality decisions rely heavily on accurate and reliable data.
The concept of cognitive dissonance in data perception—misperceiving or misinterpreting data—refers to the psychological discomfort individuals experience when their existing beliefs or attitudes clash with new information, leading them to attempt to reduce their discomfort by ignoring the conflicting data. When people experience cognitive dissonance, they attempt to dismiss or downplay the conflicting information or selectively attend to information that supports their existing beliefs, a phenomenon known as selective exposure. Cognitive dissonance can afflict those interpreting data if they allow confirmation bias to affect their perceptions and do not search for alternate (or even more plausible) explanations. Understanding cognitive dissonance is crucial when presenting data to the public, particularly in health care. By recognizing how people react to conflicting information, we can better tailor messages and strategies to promote understanding of health care data.
Another example of the disastrous use of data with questionable accuracy, reliability, and fidelity is the prosecution of interventional cardiologist Dr. Richard Paulus. From 2008 to 2024, Dr. Paulus was tried twice, spent a year in jail, and ultimately had all charges dropped when revelations about inaccurate and unreliable data in his case came to light. He was accused of the fraudulent placement of cardiac stents in cases where coronary artery narrowing was minimal and did not necessitate stent placement. His defense team successfully argued that the data presented by the government was inaccurate because the pixelation of coronary artery images was reduced by 70 percent, compromising their fidelity. It was also revealed that the 70 cases used in his trials—where he was accused of fraudulently and unnecessarily placing cardiac stents—were part of a larger sample of 1,100 cases. The remaining 1,030 cases had no questions about the appropriateness of stent placement, meaning the number of disputed cases was about 6 percent rather than 100 percent.
We live in the age of information and data, and the COVID-19 pandemic underscored the need for rapid and accurate data on disease dissemination, mortality rates, and population statistics. The utilization of artificial intelligence may amplify issues related to data fidelity and accuracy. A focus on data accuracy is paramount in further advancing the field of medicine.
Muhamad Aly Rifai is a practicing internist and psychiatrist in the Greater Lehigh Valley, Pennsylvania. He is the CEO, chief psychiatrist and internist of Blue Mountain Psychiatry. He holds the Lehigh Valley Endowed Chair of Addiction Medicine. Dr. Rifai is board-certified in internal medicine, psychiatry, addiction medicine, and psychosomatic medicine. He is a fellow of the American College of Physicians, the Academy of Psychosomatic Medicine, and the American Psychiatric Association. He is the former president of the Lehigh Valley Psychiatric Society.
He can be reached on LinkedIn, Facebook, X @muhamadalyrifai, YouTube, and his website. You can also read his Wikipedia entry and publications.
