The evolution of clinical practice, between artificial intelligence and Real World Data

The evolution of clinical practice, between artificial intelligence and Real World Data

The evolution of clinical practice

This was discussed at TheBigDate together with the leading Italian experts in digital health and high tech applications for medical scientific research: this is how the way of doing research and clinical practice change between wearable devices, healthcare big data, application frontiers and ethical issues

(photo: Luke Chesser / Unsplash) What can be done, in practice, to make the most of the potential of digital health and data collected from the real world? This is the question that was at the heart of TheBigDate, the in-depth online event promoted by Pfizer Italy and dedicated to the evolution of clinical practice through the potential of cutting-edge technologies. From which, however, it emerged first of all that having the most advanced hardware and software solutions is necessary but not sufficient: behind the paradigm change in medicine, in fact, there are cultural, methodical, regulatory, ethical and - of course - scientific issues.

"An interesting and contemporary perspective is to equip machines with the ability to perceive the world", began Fabio Moioli, Director of Consulting Services at Microsoft Italy. "A perceptive capacity that is applied in the field of security, but with concrete examples that are already operational even for blind people", who need to be involved on a social level and can do so thanks to special glasses that transform visual information in words to listen to.

During TheBigDate it was also talked about how the same technology, developed in the United Kingdom, is used for example in oncology for image recognition, using the great capacity of artificial intelligence to identify correlations and make predictions based on statistics. With the idea of ​​human-machine complementarity: "If I show an exam to both an algorithm and a doctor, the result is better than if only the doctor or only the algorithm saw it, since human beings and machines think in different ways and they make different mistakes ", continued Moioli.

And looking a little further, we arrived at the most futuristic perspectives: on the one hand, quantum computing applied to healthcare big data, from another use of synthetic DNA as a solution for data archiving, up to and including the evolution of the doctor's role, with the aforementioned complementarity between people and artificial intelligence that allows healthcare personnel to have more time to dedicate to the relationship empathic with patients.

The challenge, also reiterated by Giovanni Corrao of the University of Milan-Bicocca, seems to be precisely that of harmonizing science and technology, clinical practice with today's wide availability of data . "Today we have about 7,000 innovative drugs in development", he explained, "and the normal length of the approval process of about ten years is greatly reduced, with drugs registered after small phase 1 and 2 studies, and skipping the step 3. The implication of all this is that the importance of clinical practice data, ie referring to when the drug is on the market, is getting bigger and bigger ".

However, it is not a question of finding alternatives to evidence based medicine based on randomized scientific studies, but to complete this model with the precision medicine approach, which goes beyond the evaluation of the single therapy and also includes an eye on expenses, pointing towards the paradigm of the so-called value based medicine. "It is no longer a question of assessing whether a single drug or treatment is effective, but whether or not the entire therapeutic path is suitable for the patient's well-being," added Corrao. To do this, the answer is real-world data, the data we generate by leaving footprints in the healthcare system, to be managed through huge, well-built databases.

Data that, more recently, are starting to arrive through special high-tech devices . From an impromptu survey conducted among participating doctors, for example, it emerged that in a third of cases they take advantage of health apps, but also of smartwatches in one case out of 7, of IoT sensors in an equivalent way, and also of bracelets smart, and so on. Even if, as Corrao concluded, "the important thing is that science is good science, that the clinic is good clinic, while the tools have a relative importance because they change constantly".

Established that the artificial intelligence allows to collect, analyze and process a whole series of information that usually did not enter the databases, including images, written texts and much more, one of the key points today is in the structuring and characterization of the data itself. "The effort is to make the data adequate to the Fair paradigm, acronym for Findable, Accessible, Interoperable and Reusable," explained Riccardo Bellazzi of the University of Pavia. "To use artificial intelligence well, we need to understand what we insert into our data, and we can only do so if we model the treatment process in the correct way, thinking about which algorithms to use and what kind of evidence to obtain from them".

In other words, given that artificial intelligence relies on large data collections, it is the process of collection itself and the culture of the data that make the difference, since today it can include a series of unstructured information . "The ultimate goal of the whole process is to build a system that learns from our data, in order to extract knowledge and then move from knowledge to action. And this is what must be kept in mind when it comes to Fairness, "he added.

In addition to being great sources of data, digital health tools also have great potential as health opportunities, because they promote styles of healthy life and promote medical research through the data they collect. “Many devices can passively collect our physiological data, and the challenge is to relate them to patients' clinical outcomes,” said Eugenio Santoro, director of the medical informatics laboratory at the Irccs Mario Negri Pharmacological Research Institute.

(photo: Pixabay) The already concrete fields of application are numerous: from the studies that measure the quality of sleep and the correlation with its duration up to those that evaluate the mobility of the joints, comparing the data collected by monitoring devices with what patients themselves declare about their state of health. "Real world data studies performed using digital health tools enroll patients in much shorter times, they can enjoy data collected from different sources and on a regular basis, improving the accuracy of the studies and reducing their costs," he added. "And in future studies, randomization will increasingly take place using social media systems, with drugs shipped home, data based on patient reported outcomes and also analyzed through artificial intelligence systems".

One of the critical points reiterated is However, if data is needed to carry out the analyzes, in Italy these come from 21 different systems (between regions and autonomous provinces), which speak little among themselves and are neither interoperable nor easily accessible. "It is a situation that has emerged strongly also with the health emergency", Santoro concluded, "and we need first of all to redesign our electronic health record in a clinical key".

"There are two paths that go into parallel ", added Daniela Scaramuccia of IBM," on the one hand we must improve the way in which we collect data, and we expect the National Recovery and Resilience Plan to give a strong push in this sense, but in parallel we cannot throw away the enormous amount of data we already possess and which are structured in silos ". The great advantage we have today, as it has emerged, is that artificial intelligence allows us to process this data, bypassing that phase of manual work that made it unusable in the past.

In a broader sense, however, when it is about the evolution of clinical practice and real world data we must not forget the questions of balance, for example between artificial intelligence, ethics and regulations. "We need algorithms that are free of bias and that are explainable and understandable, but above all we need agreement and harmony between stakeholders and regulatory bodies, and that we continue to deal with these issues", concluded Scaramuccia. Finally, taking care of a further fundamental point of balance: that between the possibility of doing research and individual rights, between the usability of data to allow scientific advances and the confidentiality and privacy of people.

Medicine - 5 hours ago

Which vaccine to choose after a first dose with AstraZeneca?

adsJSCode ("nativeADV1", [[2,1]], "true", "1", "native", "read-more", "1"); Startup - 5 hours ago

NATO hunts for military startups against China and Russia

adsJSCode ("nativeADV2", [[2,1]], "true", "2" , "native", "read-more", "2"); Environment - 23 hours ago

Cremona, more famous for smog than violin


Big data Artificial intelligence Health globalData.fldTopic = "Big data, Artificial intelligence , Health "

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.

Powered by Blogger.