In an article entitled “11 clinical trials that will shape medicine in 2024” Nature Medicine illustrates the responses of 11 expert researchers who were asked to indicate which clinical trials are destined to have the greatest impact in 2024. Among the various innovations that characterize the trials, digital technologies and AI promise revolutions in the fields of diagnosis and treatment, as demonstrated by these studies on the use of Machine learning for patient triage, on an app for perinatal depression and on the use of AI in diagnosis early lung cancer.
Here is in detail what the studies involving digital technologies and artificial intelligence deal with.
App for perinatal depression
Atif Rahman: professor of child psychiatry and global mental health all’Università di Liverpool
A team of researchers led by the University of Liverpool have developed an application (THP-TA) that allows a woman without healthcare experience to provide a cognitive therapy-based intervention to women in the second or third trimester of pregnancy who suffer from major depression. This study will compare the application with the standard face-to-face version of the Thinking Healthy Program therapy delivered by health workers in rural Pakistan. This study could bring further innovations in this area by addressing the treatment gap for a common mental disorder globally.
Machine learning for patient triage
Steven Meex: head of the unit general clinical chemistry and hematology del Central Diagnostic Laboratory of Maastricht UMC, Olanda
In a recent retrospective study conducted at the Maastricht University Medical Center, a new clinical risk score, the RISKINDEX, was introduced using an AI model to predict the 31-day mortality of patients who attended an emergency room. The tool was developed based on data from 266,327 patients with 7.1 million laboratory results available. RISKINDEX has outperformed internal medicine specialists but it is unknown to what extent these AI models have beneficial value when implemented in clinical practice. MARS-ED is a prospective, multicenter, randomized, open-label, non-inferiority study. It is a pilot study of risk-scoring assistance to emergency room doctors, the objective of which is to determine the diagnostic accuracy, clinical and organizational impact of the RISKINDEX as a basis for the conduct of a randomized, multicenter and large-scale study. stairs. Prospective validation validation of AI models is rarely done but is indispensable, because implementation in clinical practice can be disappointing.
Artificial intelligence (AI) for the early diagnosis of lung cancer
David Baldwin: honorary professor of medicine alla University of Nottingham, UK
Diagnosing lung cancer early would allow for better and more effective treatment. To date, a chest x-ray (X-ray) is the first test that suggests the presence of lung cancer and, if followed promptly by a computed tomography (CT), can advance the diagnosis. In this regard, a randomized and controlled clinical trial (qXR) is underway on 150,000 patients to verify whether a deep learning algorithm applied to chest X-ray and CT reduces the time to diagnosis. The application of AI could reduce the time to diagnosis of lung cancer by up to 50%, thus bringing an immediate change in the diagnostic path of the disease.
