The Health AI Observatory publishes the Study of the Maturity of Health Artificial Intelligence in Research Centres in Catalonia

The Health AI Observatory, coordinated by the TIC Salut Social Foundation, has identified 50 artificial intelligence (AI) algorithms that are being developed or used in different health research centres in Catalonia. The Study of the Maturity of Health Artificial Intelligence in Research Centres in Cataloniacollects these algorithms and classifies them in categories such as the degree of maturity reported, the care area, medical speciality or type of procedure to which it applies.

To carry out the study, Catalan research centres involved in health were identified and a survey was sent to them. Of all the centres that answered (31 of the 39 invited), 67% stated that they are developing or using AI algorithms mainly in the field of health.

Algorithms by degree of maturity

The degree of maturity of the algorithms has been measured on a nine-point scale that describes the progress made with the technological solutions. This scale goes from Technology Readiness Level 1 (TRL 1), which is the design stage, to TRL 9 for algorithms that are already in the full market deployment stage. The different TRL levels are divided into three major environments: laboratory environment (TRL 1 – TRL 4), simulation environment (TRL 5 – TRL 6) and real environment (TRL 7 – TRL 9).

According to the data reported by the research centres, the average degree of maturity of the algorithms is at a TRL of 4.78. Half of the algorithms are in the conceptualisation and design phase at TRL 2, 3 and 4.

Regarding the breakdown by environment, the study shows that 54% of the health algorithms are in the laboratory environment, 32% in the real-world environment, and 14% in the simulation environment.

Algorithms by care area and speciality

The algorithms identified in the report are being developed or used, in some cases, in different care areas and medical specialities at the same time. 43% of the algorithms are applied in hospital care, followed by primary care with 30%, mental health care with 13%, and social health and research, both with 7%. In terms of medical specialities, medical oncology, cardiology, and endocrinology and nutrition predominate.

The study also classifies the algorithms into three main types of procedures: 72% of the algorithms identified are involved in a medical or surgical procedure, 22% in management procedures and 6% in maintenance procedures.

Periodic monitoring

The Health AI Observatory will periodically monitor the identified algorithms and will regularly update the indicators of the algorithms used both in research institutions and in Catalan Health System (SISCAT) centres on its web page. Among these two group and private initiatives, the Health AI Observatory has now identified 107 health AI algorithms.

According to Núria Abdón, the head of the Health AI Observatory, “Identifying the algorithms makes it possible to know the degree of development and implementation of AI in the field of health, share and report on good practices, and promote the adoption of this technology in Catalonia, among other benefits.”