The UPV and INCLIVA promote a project to improve diagnosis and treatment through big data and artificial intelligence techniques.
INCLIVA Health Research Institute of the Hospital Clínico de Valencia, together with the Universitat Politècnica de València (UPV), has promoted a project aimed at improving diagnosis and treatment systems in psychosis by incorporating artificial intelligence and big data techniques.
According to the head of the project, Dr. Julio Sanjuán, coordinator of the INCLIVA Psychiatry Research Group and the First Psychotic Episodes Unit of the Hospital Clínico de Valencia, this “stems from a long trajectory, of more than 20 years, of this research group looking for indicators (biological markers) that help us both in the diagnosis and prognosis in psychosis”.
The main objective, adds Sanjuán, “is to be able to predict, at the first psychotic episode, the response to treatment and the course of the disease, in order, in short, to offer the optimal treatment to each patient”.
María José Castro (VRAIN-UPV), key in the use of artificial intelligence techniques
The project takes into account two strategies to improve diagnostic and therapeutic systems in daily clinical practice. On the one hand, the analysis of follow-up data from the total sample of patients seen in the First Episode Unit, which are representative of the overall population.
And, on the other hand, the use of artificial intelligence techniques, for which María José Castro, from the Valencian Institute for Research in Artificial Intelligence (VRAIN) of the UPV, expert in data analysis, through machine learning, for the generation of a diagnostic-prognostic algorithm, has joined the research group.
Psychiatry, a pending subject of medical advances with machine learning
At present, there are very promising preliminary results from the use of this technique with functional magnetic resonance imaging, and the intention of the researchers is to apply such analysis to the whole data set, including clinical, genetic and neuroimaging data. As for the collection of such data, the project is well advanced, with the registration of those belonging to more than 200 individuals.
So far, machine learning analyses have been performed only with functional neuroimaging data, but, in a few years, it is expected that an analysis of the whole dataset will be available, for which this type of resonance imaging will be performed in both national and international hospitals. In addition, the study will be extended to include data from the patient’s medical history and gene data.
“Machine learning techniques are already transforming medicine, reducing the time needed to reach a diagnosis,” Castro points out. “New and more advanced solutions are appearing every year, especially in fields of medicine associated with tumor detection or other degenerative diseases. But these advances have not been achieved in the field of psychiatry, because there are no detectable biological markers,” he warns.
Today, deep learning techniques can analyze many more factors and cases than human specialists. To be more precise, these techniques can be used, among other applications, for genome research, drug development and medical imaging. These automated systems can learn and analyze large amounts of information and make decisions much faster than humans.
“We should move towards a joint collaboration between human experts with these automatic techniques, using the results of diagnostic aid algorithms as support systems for making medical decisions,” adds the researcher from the UPV’s VRAIN Institute.
Project members and financing
The project involves, on behalf of Dr. Sanjuán’s multidisciplinary research group, the clinical researchers Eduardo J. Aguilar and María Dolores Moltó (geneticist), as well as animal model researchers such as Juan Nàcher. Gracián García-Martí and Luis Martí-Bonmatí, from the Imaging Department of the Clínica Quirón, are also participating.
In addition, collaborations are being carried out with other Spanish research groups through CIBERSAM, and with other foreign groups through international consortiums.
The project has a total funding of 349,640 euros, from FIS 01/01/2018 – 31/12/2020 Functional magnetic resonance imaging and gene expression as predictors in first psychotic episodes (110,000 euros); PROMETEO – Investigation of biological markers and new therapeutic strategies in psychosis (179,940 euros); and INNVAL – New technique for individualized diagnosis of psychosis, based on learning (59,700 euros).
Psychosis, one of the leading causes of disability worldwide
Psychosis (schizophrenia and bipolar disorder) is one of the leading causes of disability in the world. The prevalence of psychosis is between 3 and 4%. The cost of treating psychosis in Europe is around 93.9 billion euros. Many studies have shown that early detection and early treatment not only improve clinical prognosis, but also clearly reduce overall costs.
Despite this interest, there is a worrying dissociation between research and clinical practice in psychosis research. Although important advances have been made, these have not, in general, translated into an improvement or change in diagnostic and treatment systems in clinical practice.
Therefore, research into the causes and new therapeutic strategies in mental illness is a priority for the World Health Organization, the Instituto de Salud Carlos III, and other health organizations at international, national and regional levels.
Click here for more information (spanish)