A software robot has proved to be faster than doctors at detecting side-effects during a drug treatment for cardiac arrhythmia, while also cutting unnecessarily frequent follow-up lab tests and controls, according to a study conducted at the University of Gothenburg (Gothenburg, Sweden).
The study—the findings of which are now published in the Journal of Medical Internet Research—used a proprietary guideline-based method to automate and facilitate the monitoring and evaluation of side-effects when following up amiodarone (Cordarone) treatment. Amiodarone is used to help treat atrial fibrillation (AF) and tachycardia, and requires regular blood tests due to potential side-effects on thyroid and liver function. These tests involve a completely manual process under current practices.
The purpose of the study was to develop and validate a robot prototype that is not artificial intelligence-based, and follows prevailing healthcare guidelines.
The robot was programmed to search for appropriate patients from an administrative list, check their test results against an algorithm, and present a recommended course of action for the attending cardiologist. Once the physician makes their decision, the robot adds a new entry to a waiting list to book the next blood test appointment.
The study tested the robot’s ability in parallel to the real-life treatment of patients—without the cardiologists or patients being aware of the robot’s assessments. Following this, the researchers were able to compare the outcome of the physicians’ interventions with what the robot would have contributed had it been involved.
The robot’s ability to perform routine assessments was tested against doctors’ orders in 390 situations. On average, the software robot recommended a 4.5-month interval between laboratory tests, which the researchers state is completely in line with standard practices. The doctors prescribed an average interval of 3.1 months. In addition, based on the patients’ blood tests, the robot found 12 side-effects at the first attempt, while the doctors detected eight during the initial test and the remainder during subsequent tests.
The researchers believe that the automated process—aided by a software robot and a diagnostic classification algorithm—is a technically and medically reliable option for following up amiodarone treatment and, by extension, other drug treatments as well.
“The method was shown to reduce manual work tasks, reduce the frequency of lab tests, and improve the detection of side-effects,” said Helén Sjöland (University of Gothenburg/Sahlgrenska University Hospital, Gothenburg, Sweden), a leading author for the study. “Accordingly, it would likely reduce healthcare costs while increasing the value for patients.
“The robot does what a human would otherwise do—it simply follows the rules, which are actually quite complex. As practising physicians, we’re too cautious and order some tests unnecessarily to play it safe.”
The method outlined in this study has been registered as an in-house manufactured medical device under the EU Medical Device Regulation (MDR) and is currently being implemented at the cardiology department of Sahlgrenska University Hospital.
