Research was awarded by the Brazilian Society of Cardiology and the European Innovation Congress
Research conducted at the Aterolab Laboratory at the Faculty of Medical Sciences (FCM) at Unicamp tested the use of artificial intelligence to identify patients with chronic coronary diseases and classify them as having greater or lesser risk of adverse clinical events. The model – based on algorithms – also made it possible to estimate the costs of these patients for the Unified Health System (SUS). The purpose of the study, according to Unicamp researchers, is to increase the accuracy of hospital discharge and help with medical decisions.
“When the doctor cares for a critically ill patient with acute coronary disease, he needs to decide whether the patient can be discharged, whether he needs another type of intervention or high-cost care, whether he can go to a public hospital or stay in a hospital. of high complexity. To facilitate this decision, we created some sophisticated algorithms based on equations based on the sum of patient characteristics”, explained cardiologist Andrei Sposito, coordinator of Aterolab and research advisor.
Using data from medical records of patients treated in public and private hospitals, researchers analyzed 1.089 individuals with acute coronary syndromes between 2006 and 2018 and modeled 29 possible clinical variables for cardiac patients. These variables included vital signs, coronagraphic data, diagnoses and laboratory data. For cost analyses, the researchers considered the possibility of new revascularization, cardiovascular and non-cardiovascular deaths, chronic dialysis treatment, hospitalizations and cerebrovascular accidents (CVA).
After capturing the information collected from paper or electronic medical records, the data was entered into the system and the computer was “trained” to simulate possible scenarios for patients with chronic coronary disease. Then, with each new piece of data collected during clinical or outpatient care and entered into the system, the computer recalibrates itself.
“The computer learns from each result and informs the doctor of the best clinical course of action for the patient. From this work, we were able to predict 92% of the clinical events that a patient may experience. We also identified patients at high risk for acute coronary syndromes whose cost was almost five times higher. The application of artificial intelligence to these patients could represent savings of almost R$50 million per year for Brazilian hospitals. Furthermore, we will be sure that the most seriously ill patients will be receiving more accurate treatment”, said cardiologist Luiz Sérgio F. de Carvalho, researcher at Aterolab and main author of the study.
The results of the research entitled "Artificial intelligence predicts the risk of new events after acute coronary syndromes, identifies high-cost individuals and those with a high burden of uncontrolled risk factors" was presented during the 74th Brazilian Congress of Cardiology, held in September Porto Alegre and won the award for best free theme. The study also included the participation of Silvio Gioppato, Marta Duran Fernandez, Bernardo Carvalho Trindade, José Carlos Quináglia e Silva and Sandra Ávila.
After presenting the research during the Congress, the Brazilian Society of Cardiology (SBC) invited researchers from FCM's Aterolab to create, within SBC, a committee to develop an artificial intelligence project with the same characteristics developed at Unicamp and distribute it lo, nationally, for public hospitals in the next two years.
“Our study showed that the use of artificial intelligence improves the effectiveness of care. We are able to identify who is more likely to die, return to the hospital or cost more. We were able to identify 10% of patients who consume more than 30% of the resources spent. We can prevent all of this from happening using this new technology”, reinforced Luiz Sérgio.
The research was presented in early November during the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) Europe 2019, in Denmark, and came first for best work. ISPOR is the most important European innovation congress.
“The United States is encouraging hospitals that adopt electronic medical records that interface with the artificial intelligence system. In no time, they will be running it flawlessly. In Brazil, some private hospitals are pursuing this. Those who deal with medical management understand that there is no way out. This is the best way to work. For the patient, it is a safe system. For the hospital it is cheaper, as it optimizes actions. It’s the future that has arrived”, concludes Sposito.