- November 22, 2022
To better identify individuals at risk, some psychiatrists and institutions rely on digital technologies. In the line of fire: facilitating diagnoses and access to care. “No health without Mental Health “recalls the WHO. So to alleviate our mental disorders in the context of degradation of public hospitals and the health sector which is hitting psychiatric services hard, some are turning to new technologies. This is the case of Stéphane Mouchabac, a psychiatrist at Saint Antoine Hospital in Paris. He is part of the research unit of the Institut du Cerveau in Paris and co-directs the e-health section of the French Association of Biological Psychiatry and Neuropsychopharmacology. Its objective: to study the impact of new technologies on psychiatric care and develop — in collaboration with teams specializing in artificial intelligence — therapeutic digital tools for patients and professionals.
Why in psychiatry use technology?
Stephane Mouchabac: We have recently gone through periods that have affected the most vulnerable (attacks, health crises, geopolitical tensions, etc.). What happens between two consultations if patients need help, how can they be accompanied in an emergency if they are alone? It should be borne in mind that the management, which is always complex, of mental illness comes up against a lack of means. Access to care is sometimes limited in terms of supply, psychotherapies can have a cost that cannot be assumed by all. Moreover, the brain, a complex organ, is far from having revealed all its secrets. Digital technologies can appear as an interesting solution, even if they are not the only alternative. Note that in 2020 there were more than 10 billion connected objects, 80% of which were related to health. Digital psychiatry can, in my opinion, contribute to improving current care by allowing easier access to a greater number of people, and perhaps in a less stigmatizing way. It also brings new therapeutic modalities via applications and makes it possible to produce new scientific and medical knowledge using algorithms. It is in this context that some believe that we can better detect and prevent the occurrence of psychiatric symptoms. In effect, many parameters can be digitally collected in real-time, true behavioral markers capable of characterizing the “digital signature of a pathology “.
What solutions are in place so far? What do the applications and algorithms offer?
S.M.: We are still at the beginning! Clearly, many of the applications identified are still research. On the market, we mainly find “digital therapies” centered on specific pathologies. The United States structured e-health research several years ago, offering solutions approved by the Food and Drug Administration in the field of depression or addiction. In Europe, health policies have understood the issues, and digital doctrines have been proposed to promote the emergence of these tools while ensuring their reliability and usefulness. The designers can be institutions (research units such as INSERM or CNRS in France), start-ups, or even pharmaceutical companies that invest in these technologies. But the processes are slow; the different actors do not always have the same timeframe: research takes time, while the markets evolve quickly…
How can our smartphones help detect suicidal risks?
S.M.: The smartphone is an ideal tool for this type of model. It makes it possible to process information from internal sensors reflecting our level of energy and activity (GPS, accelerometer, etc.), peripheral indicators (biological parameters), or even to analyze the nature of our communications (processing language and emotional analysis, time spent communicating, number of contacts). From these data, and if large inter-individual databases are available, the algorithms can test hundreds of relevant configurations that can characterize a pathology. They also make it possible to compare behavioral variations in the same subject that may be linked to relapse. The content of social networks is rich in information, GAFAM knows this only too well… But we can also use them to, for example, identify a person at risk. Algorithms can automatically determine the digital signature of a suicidal tendency from the analysis of its communications.
Some algorithms could also detect schizophrenic symptoms by analyzing a few tweets. How?
S.M: Experimental work published in 2015 looked at social media content to see if an illness such as schizophrenia could be detected among Twitter users. From parameters as simple as the number of contacts, the use of emoticons, the time of publication, and the delay between two tweets, parameters combined with an automated analysis of the language content, it was possible to produce an algorithm automatic called “supervised”. It is an algorithm built from the decryption of content whose diagnosis of the authors is already known. On this basis, the software identifies differences in these parameters (time of publication, etc.) between subjects suffering from schizophrenia and non-carrier subjects. The algorithm therefore makes it possible to create classification “rules” making it possible to assign a category (schizophrenia/non-schizophrenia) to new undiagnosed subjects based on their tweets. This model has a grading accuracy of 92%.
How and by whom can this type of solution be used?
S.M.: For the creators of the algorithm, the latter would make it possible to directly affect undiagnosed subjects, to help doctors, and to destigmatize the disease. More generally, the “tech gurus” today consider that it is difficult to predict the use of technologies, as knowledge and practices evolve rapidly. Some technologies are called “emerging”, others “disruptive”. Still, others remain “immature”, which is mostly the case in the field of digital psychiatry. In France, the agencies seek to promote innovation and the national deployment of digital health platforms, to allow professionals to access digital communication tools that will be carefully selected, referenced, and validated.
What are the risks associated with these new practices?
S.M.: Our society no longer tolerates uncertainty. He needs predictions, and above all that these are correct, even when it is simply a question of the weather. However, the exercise of psychiatry is based on complex and polyfactorial models… If the digital and its predictions are promising, there is always a margin of error, which at the individual level is not acceptable. We will have to learn to manage this new form of uncertainty. In oncology, tumors are detected at such early stages that one sometimes wonders about the relevance of medical care, because doctors believe that our immune system can “take care of them” effectively. Considering that certain digital markers predict a pathological event, isn’t there a risk in medicalizing behaviors too early, when they would have been managed by the subject or his environment? However, algorithms are everywhere, we use them on a daily basis: it would be a shame not to exploit their potential for the treatment of mental suffering. However, this calls for a good dose of ethical reflection and good pragmatism: the shift towards “ Big Brother ” Where ” Minority Report can happen quickly! This is why we are campaigning for the direct involvement of users, so that caregivers are also involved and that public research has the means to participate in this revolution.
- AI-Driven Agriculture
- Non-Technical Introductory AI Courses for Enthusiasts
- Microsoft’s Xiaoice AI Project
- AI-Powered Drones Outperform Human Pilots in Obstacle Course
- AI Algorithms for Game AI
- AI in Financial Services
- Stability AI Introduces Cutting-Edge ‘Stable Audio’ Model for Precise Audio Generation
- Stanford University’s Online Artificial Intelligence Lab Courses
- How AI is Revolutionizing DNA Testing
- A Fusion of Domain-Specific Assistance and Robust Language Models in an Embodied Conversational Agent
Get regular updates on data science, artificial intelligence, machine