"From artificial intelligence the answer to the healthcare of the future, even in Sardinia"

A study that starts from Sardinia, and which aims to underline the importance of an approach to medicine that is increasingly global and at the same time customized to the real needs of patient care.

An approach that sees the figure of the doctor, with his academic preparation and his own wealth of knowledge and experience, as an essential and pivotal element, but with the basic idea of the importance, in terms of results and benefits, of the intervention at a later stage of database-based artificial intelligence.

“Evolution of Clinical Medicine: From Expert Opinion to Artificial Intelligence” is the title of the publication, reported in 2021 in the prestigious Journal of Translation Critical Medicine.

Among the authors many Sardinians, such as the nephrologist from Cagliari Antonio Barracca , the developer and expert in "machine learning" Mauro Contini , the electronic engineer and developer Stefano Ledda , the electronic engineer and expert in "computer science" Gianmaria Mancosu and the 'senior engineer of CRS4 Giovanni Pintore . Next to them, Kianoush Kashani , Nephrologist of the Mayo Clinic of Rochester in Minnesota, and Claudio Ronco , Director of the International Renal Research Institute of Vicenza.

Antonio Barracca, also founder of abcGO !, a start-up present in 45 countries with 9 Apps that contain more than 100 applications to aid clinical practice, explains it to L'Unione Sarda.

“This is a study that already in the professional profile of the authors - doctors, statisticians, developers and technology experts - highlights its main objective: that is to focus on how today the figure of the doctor must increasingly be interconnected with other professions. The starting point is always the professional, who in front of a patient must be able to use knowledge and skills to arrive at a diagnosis as accurate as possible ".

An example?

"For example: when a patient arrives in the emergency room, the doctor must provide a first and timely response. The role of the machine, appropriately trained by specific professionals, comes at a later stage, for example by crossing millions of data that a human being , for obvious reasons, he cannot do it, and by looking for similar cases to learn about the therapies implemented in the world and the clinical results achieved ".

A real IT revolution available to doctors ...

“Nowadays there is a lot of information that is collected through different channels. Most come from Internet of Things (IoT) systems connected to telemedicine devices that constantly monitor the patient. But they can also come from the various imaging of x-rays or CT scans. The goal is to transform all this wealth of information into knowledge. Think of the famous Apple Watch, which constantly tracks the patient's heartbeat and other vital signs. Making this information available to an expert, and doing it quickly, can mean early and accurate diagnoses. Or even calculate the risk, for each patient, of suffering from a certain pathology in the future. This would truly be a revolution ”.

In the study, “data mining” is given great emphasis: why is it so important and what do you need to know about specific professional figures?

"'Data mining', literally from the English 'data extraction', is the starting point, that is the set of techniques and methodologies that have as their object the extraction of useful information from large amounts of data, the so-called 'big data', information that 'machine learning' must then examine. The techniques and strategies applied to 'data mining' operations are largely automated and consist of specific software and algorithms suitable for a single purpose. But they require particular computer skills and master's studies, and hence the need for doctors to be increasingly supported by specific professional figures ".

Can you now explain better what is "machine learning", another central aspect of the study, and what is behind the automatic learning of machines?

“Artificial intelligence uses statistical methods to progressively improve an algorithm's ability to identify reference patterns in data analysis. Machine learning, in practice, is therefore a model that learns from examples. For each task, examples are provided to machine learning in the form of inputs or 'classifications' and outputs or 'labels'. Let's take for example a histological exam read by a pathologist, digitized and converted into classifications (the set of all the pixels that make up the exam) and into labels (the information that classifies the type of pathology present). Using an algorithm to learn from observations, you can determine how to run a map that combines classifications with labels to create a generalizable model. Therefore, each new model will be performed correctly even on new histological examinations never seen before by a pathologist ".

Are there any virtuous examples to look at in terms of artificial intelligence applied to medicine?

“Doctor at hand” is an application launched in 2017 and today very widespread in the United Kingdom: it was born from the collaboration between Babylon Health, a health services company, and the National Health Service (NHS). Through the App, the patient can access a triage system based on a chatbot that analyzes the symptoms reported and proposes some answers, for example booking a video appointment with a family doctor or, in selected cases, going to the emergency room. Citizens can also directly access consultations, available 24 hours a day, usually within 2 hours of the request. The consultation can be followed by a pharmacological prescription, a treatment advice or an indication to a specialist consultation.

The target?

"Babylon Health's mission is to reduce the workload of doctors and hospital services, especially emergency rooms, and provide a better service to citizens, in order to make the health service accessible, affordable and, in fact, 'within reach but no".

Another example?

"Ping An Good Doctor ', the giant led by Fang Weihao and the most important Chinese telemedicine operator, with over 300 million users, one fifth of the Dragon's population. It offers digital medical services and an e-commerce platform specialized in health care with professionals available at all times, facilitated by artificial intelligence applications. And not only: it has also created instant clinics where it is possible, 24 hours a day, to receive immediate and punctual diagnoses.

Many of the people who rely on this app are people who live in rural areas and therefore lack adequate services. And the most important aspect of the application is that behind it is the work of hundreds of scientists who have produced a series of precise guidelines: the machine works, therefore, on the basis of information that the professionals have provided. The diagnosis is not then an automatic, but takes place on the basis of the scientific literature ”.

In Sardinia there is a real shortage of white coats. Do you think that tools of this type could find effective applications on the island?

“In Sardinia, the first point to be addressed, in this sense, is the issue of digitalization and computer literacy among all segments of the population, which still sees a long way to go. There is also a great shortage of white coats. And a need to reorganize health care at best. Let's start from a key point: it is said that there are 1400 doctors missing on the island, a fact which, however, is considered organic from the past. So why not hypothesize an update of the workforce, perhaps relieving the most experienced doctors from shifts and night guards and thus giving them the time to study and share all the data available and contained, for example, in the medical records to be digitized? Then there is the issue that in Sardinia there are many small hospitals with diagnostic services, but there are no radiologists. Why not provide a single diagnostic center in which to process the various images acquired in the various locations? It would be a response to the need for more timely and most likely more accurate diagnoses ”.

Are there, even in this area, models which the island could draw inspiration from?

“I'll give you the example of Zebra Medical Vision, an Israeli startup that uses artificial intelligence technology to help read CT scan data. The company, which has received grants from the Israeli Innovation Authority, works to detect early signs of breast cancer and osteoporosis, analyzing information and producing medical reports with 90% accuracy. In the United States, for some time now, the reports of some diagnostic imaging tests have been processed, to speed up time, by experts connected remotely from Bangalore and dedicated solely to this activity ".

New medical culture, increasingly marked use of artificial intelligence and big data, and more and more people connected to the web: what other ingredients for tomorrow's healthcare?

“First of all, the inclusion of specific professionals, coming out of universities, who increasingly respond to this 'integrated' approach with machines: it is useful for doctors to be supported by experts in statistics, engineering and IT. And then the ability to systematize data, something that happens too rarely in Italy today: electronic medical records are almost never used, in hospitals often the patient file is updated only after discharge and with the only indication of the final diagnosis. Imagine a system in which doctors receive information about a patient who has had a stroke while still in an ambulance. Or that the data of thousands, millions of people can be used by a doctor to identify the best treatments for a patient with a particular form of cancer or to predict, as Covid teaches, the evolution of an epidemic. These are just simple examples of what digital medicine and the appropriate use of big data can already offer today. With an ethical clarification to make: the machine never makes the diagnosis, there must always be a medical manager who processes what the machine suggests. No automatism, therefore, but an important help from the machines to always do better by minimizing the possibility of error ".