The software can't catch everything – Blue Springs Examiner

In 1968, Stanley Kubrick released a movie that changed the landscape of science fiction. It was “2001: A Space Odyssey.”
One of the principal characters in the movie is a computer named Hal 9000, artificial intelligence that had the capacity to experience feelings and sensations. Hal controlled the systems of the Discovery One spacecraft and interacted with the ship’s astronaut crew. The computer engaged in deception and murder before it was finally destroyed. The movie painted a futuristic world in which computers had the capacity to destroy man.
Fifty-three years later, the capacity of artificial intelligence is mind-boggling. While computers don’t have the capacity to experience feelings and sensations, they do have the capacity to essentially replace our brains.
Approximately 11 years after Hal 9000 was unleashed on the cinematic world, Cerner was founded by three men from Kansas City, and it’s now one of the largest electronic medical record companies in the world and employs several thousand people in the Kansas City area. Its electronic medical record software can be found in many hospitals in the metropolitan area. A competitor is Epic, and its software is also used in many hospitals in the area.
I am familiar with both information systems as I spend a considerable amount of time in my practice reviewing medical records. Electronic records have saved me a lot of time and headaches. In the old days, we had handwritten records and had to decipher what nurses and doctors had written in the record. Electronic records are well organized and easy to access.
I can’t say that I prefer one provider over the other. They are organized differently, but it is much easier to review electronic records. I know the creators of electronic medical records did not invent this software to make my job easier, but it has that effect. They are time savers for health-care providers and make medical care more efficient.
A huge advantage of digital records is that I can perform word searches and find certain parts of a record easily. Optimization is a word that was not in my vocabulary 10 years ago. Optimization allows me to perform word searches rather than review hundreds if not thousands of pages of records looking for certain words and phrases.
While I believe electronic medical records have improved medical care, I have also seen some limitations. Often times there are “drop down” menus that enable the nurse or doctor to choose options. The same record can be repeated and the record defaults to the previous entry, so the nurse or physician must manually change the record if the sign or symptom of the patient changes. So, the nurse or doctor must be careful.
Another aspect of electronic medical records is that the software can issue alerts to health-care providers. I have seen it in cases of possible sepsis. Sepsis is a medical condition caused when an infection triggers a chain reaction throughout your body. Without timely treatment, sepsis can rapidly lead to tissue damage, organ failure and even death. It is considered a life-threatening medical emergency.
Epic has a sepsis prediction model in its software that helps to identify patients with sepsis. It is based on the use of algorithms. An algorithm is a process or set of rules to be followed in calculations or problem-solving. A recipe is an example of an algorithm because if you follow the rules you have a tasty meal.
In medicine algorithms are used in all kinds of situations, especially in diagnosis of medical conditions. In sepsis, certain signs and symptoms lead to the diagnosis, and the sepsis prediction model helps lead the health-care provider to a diagnosis. The computer looks at laboratory records and other parts of the medical record to give an alert for sepsis.
In June of this year JAMA Internal Medicine, a publication of the prestigious American Medical Association, reviewed a study of 27,697 patients aged 18 years or older admitted to the academic health system of the University of Michigan, analyzing the accuracy of the Epic Sepsis Model (ESM), a sepsis prediction model. Under this model system, the score was calculated every 15 minutes, analyzing certain data such as temperature, heart rate, respiration rate, white blood cell count, platelet count and serum lactate.
In this study, the ESM identified 183 of 2552 patients who did not receive timely antibiotic treatment, but tragically, it also failed to identify 1709 patients (67 percent) with sepsis despite generating alerts of a score of 6 or higher. The study concluded that the scoring system had poor discrimination and calibration in predicting the onset of sepsis. According to the authors of the study, “the widespread adoption of the ESM despite its poor performance raises fundamental concerns about sepsis management on a national level.” The authors noted that hundreds of hospitals have used these models with loose federal regulation.
Too err is human, but to allow computers to make medical decisions about diagnosis and treatment may also lead to errors. There is no substitute for sound medical judgment. I just hope that this Michigan study is not the tip of the iceberg. We can’t allow computers to repeat the sins of Hal 9000.
Bob Buckley is an attorney in Independence. Email him at