Diagnosis and treatment treatments that are increasingly early, predictive and personalized thanks to the application of artificial intelligence in healthcare.
A revolutionary process accelerated by the Covid-19 health emergency that is showing the benefits of artificial intelligence in personalized predictive medicine.
"Today, artificial intelligence algorithms are able to extract a large number of quantitative characteristics from medical images, often invisible to the naked eye, and to associate them with clinical data to inform doctors about the diagnosis, prognosis and response to therapies". Has explained Isabella Castiglioni, Full Professor of Medical Physics and Machine Learning at the University of Milan-Bicocca, who attended the workshop "Intelligent Imaging: Beyond the Future and Back to Mind" organized by the Bracco Group at the Italian Pavilion of Expo 2020 Dubai on the occasion of the "Health & Wellness Week" held from January 27 to February 2.
Artificial intelligence for personalized predictive medicine
Professor Isabella Castiglioni is co-founder and Honorary President of DeepTrace Technology, a spin-off of University School of Advanced Studies IUSS Pavia, which combines artificial intelligence with diagnostic imaging in the path of personalized predictive medicine of certain neurodegenerative diseases such as Alzheimer's disease, cancers, cardiovascular diseases and Covid-19.
To find out more, we interviewed the professor Isabella Castiglioni.
What is the role of AI in personalized predictive medicine?
Scientific evidence shows that each individual has a different risk of developing a disease, progressing into the disease and responding to therapy. The factors that determine this risk are mainly related to genetics and the lifestyle of each of us, that is to our phenotype. Predicting the individual risk is a very complex and often impossible task for doctors. AI is proving to be a key tool for personalized predictive medicine thanks to data that can provide a variety and amount of information useful for identifying different disease phenotypes and optimizing each patient's prevention, diagnosis, prognosis and therapy. .
During his speech at the forum "Intelligent Imaging: Beyond the Future and Back to Mind" showed some cases of application of artificial intelligence involving diagnostic imaging, a technology of the Italian Diagnostic Center, capable of predicting the highest risk for some subjects of developing diseases, such as breast cancer and Alzheimer's disease. How?
The predictive ability of artificial intelligence is now proven by many research studies. However, in order to benefit both patients and doctors, research applied to AI must be technologically transferred into a medical device that complies with international directives. AI medical devices certainly offer unique opportunities to personalize screening, diagnosis and treatment treatment but they have particularly complex aspects that must be managed with competence both in terms of technologies and standards.
The examples I told at Expo 2020 Dubai show that it is possible to train an AI system to read the first signs of Alzheimer's disease from brain MRI images and clinical data, to give an opinion on the malignancy of suspicious breast lesions, to identify anomalies of the cardiovascular system which can represent increased risks of severe diseases.
Technology transfer: from research to medical devices.
A complex step that the Italian Diagnostic Center, in collaboration with the start-up DeepTrace Technologies, has made and "can today tell about its experience relating to the" memory clinic ", a path of personalized medicine predicting neurodegenerative diseases that assists the neurologist with informed decision of a CE marked artificial intelligence medical device - Trace4AD ”, underlined Professor Castiglioni.
DeepTrace Technologies
DeepTrace Technologies is a start-up, founded by Isabella Castiglioni and Christian Salvatore, researcher at the Institute of Higher Studies of Pavia (of which the company is a spin-off) to develop devices doctors aimed at improving the diagnosis and prognosis of some neurodegenerative diseases such as Alzheimer's disease, cancer, cardiovascular disease and Covid-19.
Based on analytics based on AI data, DeepTrace Technologies provides solutions that allow you to anticipate diagnosis, predict prognosis and make personalized predictive medicine pathways more efficient.
These technologies are non-invasive, scalable, replicable, sustainable and support user involvement in decision management by providing understandable results. A software medical device is delivered to healthcare professionals with a highly competitive time-to-market.
How was DeepTrace Technologies born?
I wanted to reach society, that data and information had a concrete impact on patients and doctors, I wanted the best results of scientific research to come out of academic journals and arrive in hospitals.
This idea catalyzed the enthusiasm of some very young researchers in my research laboratory, who decided to stop in Italy and start this business project with me; one of them today is the CEO of the company. Training an AI system to predict a patient's prognosis is like seeing your child walk alone, an immense joy.
CE marking tool by DeepTrace Technologies
The tools developed by the startup DeepTrace Technologies, which have received CE certification, are: TRACE4AD ™ and TRACE4OC ™.
Then there is an AI platform for scientific research, Trace4Research, which includes hundreds and hundreds of AI systems that challenge each other in training new models of predictive medicine.
Professor Castiglioni, can you tell me about these tools, the results achieved to date and the importance of having obtained the CE marking?
CE marking is very important because what we have developed are medical devices that guarantee a whole series of crucial steps concerning not only the value of the artificial intelligence algorithm, but also reliability, robustness, traceability and suitability of the software for the treatment of health data. Ultimately, devices incorporating AI systems must be understandable and transparent to users.
At the moment the tools that are CE medical devices are: TRACE4AD ™, a vertical tool of our proprietary DeepTrace platform, called TRACE4 ™, which uses the artificial intelligence tool to identify and characterize subjects with mild cognitive impairment who will progress towards the disease Alzheimer's disease, all through a simple automatic reading of the brain magnetic resonance and cognitive measures of the patient following a neuropsychological assessment.
Another vertical tool that also derives from the proprietary artificial intelligence platform is TRACE4OC ™ which is able to select women at high risk of ovarian cancer by a simple automatic reading of the transvaginal ultrasound of the suspected ovarian masses and the serum CA 125 level of the woman.
Additionally, there is another AI platform dedicated to scientific research. This is Trace4Research, which includes many artificial intelligence systems that challenge each other to train new models of predictive medicine in order to obtain the model that appears to have the highest performance with respect to certain diagnostic or prognostic tasks defined by the researcher-user. She is able to write the scientific manuscript by herself with the description of the materials and methods used to train the best model, and the results obtained.
It is a tool we are particularly proud of because it was designed for medical researchers, to break down the barriers of AI technologies that are nowadays mainly used by researchers in the world of information technology, engineering, physics and mathematics.
During the Covid-19 pandemic you "trained" the "DeepTrace" artificial intelligence platform to identify and recognize the characteristics that differentiate the X-ray images of the lungs of subjects affected by coronavirus, compared to the lungs of subjects with similar symptoms but not with Covid-19 disease. What is it about? What is the advantage of this method?
During the first wave of the Covid-19 pandemic with the arrival of many pneumonia patients in the hospital, the hospitals of the Lombardy Region San Gerardo di Monza and the Policlinico San Donato, asked us to use our AI platform to train a artificial intelligence system to recognize patients with interstitial pneumonia.
These were the first weeks of the pandemic, in which molecular analyzes took days to give the result and often patients were initially negative and became positive only in subsequent tests, so this tool was extremely useful for deciding what to do when a symptomatic person arrived at the emergency room.
Later, with the arrival of molecular swabs, we trained the system to recognize the difference between Sar-CoV-2 virus interstitial pneumonia and pneumonia caused by other types of viruses and bacteria. This tool can be very useful when the virus, in its different and new variants, will present itself with a prevalence similar to other common viruses.
What are the developments for the future?
Thanks to the platform we are able to continue this innovative development with other medical devices.
We intend to grow to become, both in Italy and in Europe, a company providing new solutions for predictive personalized medicine based on artificial intelligence. We are working on a series of medical devices that will improve the quality of life of patients with cancer, neurodegenerative, cardiovascular and musculoskeletal diseases.
For example, we have recently completed an AI-based device capable of classifying the risk of a suspicious mass to a breast ultrasound that will support the physician in referring the patient to a short-term follow-up or a mass biopsy.
In addition, we are also working on a device that improves the estimation of this risk through the reading of the mammogram and the prediction of the woman's risk related to breast density.
You have managed to bring research closer to businesses. What is your message to all women who want to undertake scientific paths from which new businesses can be born?
Thank you for the question that touches me personally. I think it is a virtuous path where research helps to create an innovative company, but in my opinion it can make a difference. It is not easy because it means working hard to maintain a high level of activities such as research, carried out with scientific independence, then made available for the development of new products and services, compliant with regulations, to be placed on the market.
The woman in the field of research plays a fundamental role because in her DNA she has the ability to carry out multi-tasking tasks in all areas of life. It takes a lot of determination and an organizational ability to carry out parallel tasks. But women are characterized precisely by these characteristics. Therefore, my message to women who want to undertake scientific paths from which to create new businesses is to use these skills and have a clear goal to reach.
From personal experience I can say that being the generator of a business path is an immense satisfaction that repays the great work and sacrifices made by a large margin.
English (UK) 
Italiano