Just published in the journal Internal and Emergency Medicine a commentary that illustrates the potential of machine learning in analyzing and processing large amounts of clinical and laboratory information, using Covid-19 as a model. The authors conclude that the approach may help doctors and healthcare professionals to identify and implement the best therapeutic treatment in real-time thus radically revolutionizing patient management.
Catania, 14 September 2022 – Artificial intelligence (AI) encloses in the same definition different applications and uses in every field of research, including medical science. AI uses complex computer systems capable of processing as much data as possible in the shortest possible time, mirroring human intelligence.
Machine learning, as part of AI science, models patterns in time-series data, through algorithms that make machines able to learn by the experience. Exploiting the possibilities of machine learning, a group of Spanish researchers of the University of Castilla – La Mancha, led by prof. Jorge Mateo Sotos, processed the clinical data of patients hospitalized for Covid-19 during the first pandemic wave, in 2020. The researchers studied whether the application of algorithms could help in recognizing predictors of mortality from clinical and laboratory parameters in patients with Covid-19.
The spanish study, also published in Internal and Emergency Medicine, was scrutinized by experts of the CoEHAR (Center of Excellence for the acceleration of Harm Reduction) at the University of Catania, in collaboration with researchers of the Bicocca University of Milan. In their commentary Pietro Ferrara, Sebastiano Battiato and Riccardo Polosa provide an analytical viewpoint of the results obtained by the Spanish colleagues, discussing new scenarios for real-time analysis in medical practice, and concluding that the application of machine learning in healthcare could increase the possibility of detecting crucial aspects of complex pathologies, helping doctors to develop effective treatment lines.
According to prof. Polosa: “In spite of the challenges posed by dataset quality, exclusion of social variables, and heterogeneity in healthcare services across regions and countries, this innovative technology has potential to improve health practices and assist health care practitioners in taking the right steps towards better prevention and cure”.
Although obvious limitations to the study emerged, mostly due to the small patients’ sample and the emergency conditions in which researchers operated owing to the disruptions in health facilities caused by Covid-19, as well as changing in disease epidemiology due to swift massive vaccinations and the emerging of Sars-Cov-2 variants: “The work by Mateo-Sotos’ and colleagues is faced by limitations due to small sample and data quality issues due to the very changing nature of COVID, but nonetheless it may provide support to the implementation of AI systems in medical research and clinical practice” – according to Dr. Pietro Ferrara, medical epidemiologist and first author of the article.
To reiterate the potential of using artificial intelligence in the healthcare system and in the medical field, prof. Sebastiano Battiato, Department of Mathematics and Computer Sciences of the University of Catania, author of the study, said: “Every day, the use of artificial intelligence makes it possible to achieve progresses that were even unimaginable only 20 years ago in any field of science”.