Algorithms for voice diagnosis and simulators for training surgeons

Developing artificial intelligence-based tools to help doctors diagnose and assess a wide range of pathologies by analysing patients' voices. This is the goal of the TALIA project on which Gpi, the Trentino-based digital innovation giant in the health sector (433 million in revenues by 2023, 7,600 employees and 9,000 customers in 70 countries) is working in collaboration with the University of Macerata, Turin Polytechnic and the University of Verona. Financed in March by Mimit with EUR 2 million, more than half of which was earmarked precisely for the Trentino group, the project starts with Talking About, a deep learning algorithm that assists in the screening of post-partum depression developed by Gpi and already in use in the facilities that have adopted its telemedicine platform (three regions, an autonomous province, and various ASLs). Various neurodegenerative, cardiovascular and emotional health-related diseases are instead the focus of the TALIA project, which will develop technologies based on the use of intelligent voice biomarkers. 'Voice is a systemic feature of our organism, like temperature, for example,' explains Antonio Colangelo, Gpi's R&D director. 'During the pandemic, family doctors had learned to strongly suspect Covid by talking to patients on the phone, because the infection produced a particularly characteristic pattern. The advantage that AI offers is that signals that are not obvious to humans, if persistent and observable in mathematical form, can be recognised with very high accuracy by the algorithm. These are very powerful tools to support doctors' actions.

Although he knows it very well, the director of the Department of Medical, Surgical and Health Sciences at the University of Trieste (and practising surgeon) Nicolò de Manzini, cannot help but think about the radical change that the Centre for Medical Simulation and Advanced Training (CSMAA) represents compared to his formative years. "Being able to learn how to insert a catheter or give assisted ventilation to an infant on smart mannequins, which indicate if they are feeling ill or if the manoeuvre is not being performed effectively, is a huge benefit for the students," says de Manzini. The case of the CSMAA, in the heart of the Cattinara hospital in Trieste, is an excellent example of how AI technologies can also be crucial in the training of new doctors and nurses. More than a thousand students enrolled in various degree courses at the University of Trieste, as well as dozens of postgraduates, use the state-of-the-art equipment at the CSMAA. These include four life-size mannequins, two adult and two paediatric, that perfectly reproduce the functioning and reactions of the human body to medical, surgical and pharmacological interventions, as well as HAL s5301, the humanoid patient with artificial intelligence and robotic limbs that constitutes the most advanced interdisciplinary simulator in the world. Among other things, the CSMAA has also recently equipped itself with two simulators for laparoscopic surgery. "One of the most complex aspects in laparoscopy is suturing," notes de Manzini, "but these simulators, which also do automated analysis of student performance and progress, are a valuable training tool, to say the least.

The RIS-PACS project for the digitisation and use of artificial intelligence in the Veneto Region's healthcare sector has begun, thanks to a total investment of around EUR 100 million from the PNRR funds. The aim is to optimise the management and administration of the various cogs of the healthcare machine, to the benefit of healthcare personnel and patients. This is an intervention that is as necessary as it is complex to implement, which requires work on several fronts, first and foremost to overcome the current, excessive fragmentation with a new architecture of the regional information systems, which will favour interoperability and the exploitation of data throughout the region. This will make it possible, for example, to unify and manage images and clinical content obtained from different specialities and facilities, and to create a single patient record to support clinical decision-making, facilitating more precise diagnosis. The project will cover all regional healthcare companies and is to be completed in 2025, as required by the NRP.