Podcast Summary: Les Clés – "L’IA pourrait-elle devenir votre médecin ?" (3/4)

Host: Arnaud Ruyssen
Guests: Dr. Brouillère, Giovanni Briganti, Nathalie Grandjean
Date: October 22, 2025
Duration: ~30 min

Overview: Can AI Become Your Doctor?

This episode of Les Clés, hosted by Arnaud Ruyssen, explores the rapid integration of artificial intelligence (AI) into the medical field. It investigates how AI is transforming diagnostics, patient care, research, and the broader healthcare ecosystem. Through expert interviews and a field report, the episode weighs the immense opportunities AI offers against the complex ethical, social, and systemic challenges it raises.

Key Discussion Points & Insights

1. Concrete Uses of AI in Radiology

Reportage by Sarah Poussey featuring Dr. Brouillère

• AI Already at Work: In Liège’s radiology department, AI technologies adapt MRI and scanner settings automatically to each patient, reducing radiation and optimizing contrast agent doses ([01:31]–[02:39]).
• Diagnostics Enhancement: AI servers help identify potential issues like lung nodules or fractures, flagging cases for radiologists’ attention. The combination of human expertise and AI results in better diagnostic accuracy ([02:54]–[04:05]).
• Urgency Detection: AI flags exams needing urgent review (e.g., hemorrhage, stroke), sending notifications directly to medical staff, enhancing patient outcomes ([04:16]).

Quote:
"On va avoir une optimisation du contraste pour injecter le moins de contraste possible au patient."
— Dr. Brouillère ([01:55])

• Limitations & Risks: AI struggles with specific cases (e.g., spotting nodules over 2 cm, false positives). Dr. Brouillère highlights the importance of clinicians understanding these tools and their limitations ([05:19]–[06:01]).

Quote:
"Le plus gros problème de l’intelligence artificielle actuellement, c’est probablement de savoir comment elle fonctionne... Les vendeurs ne dévoilent pas leurs algorithmes."
— Dr. Brouillère ([05:19])

• Human–AI Collaboration: AI is not ready to replace radiologists but enhances their work. The perfect scenario is their collaboration ([06:10]).

Quote:
"On a montré qu’il n’y avait pas 100% de réussite ni d’un côté de l’IA ni du côté des radiologues... c’est mieux de faire les deux."
— Dr. Brouillère ([06:10])

2. AI’s Growing Role Across Medicine

Interview with Giovanni Briganti, Professor at UMons/Lumons, Psychiatrist and AI Chair

• Ubiquity in Medicine: Every branch of medicine is touched by AI, from diagnosis to drug development. The general public already senses its impact will be greatest here ([07:13]).
• Diagnosis and Prediction: AI enables earlier, more accurate diagnoses and supports remote monitoring and risk prediction—especially valuable for isolated individuals or chronic conditions ([07:57], [08:55]).

Example:
Creating "digital twins" for elderly patients to monitor health in real-time and predict adverse events, like cardiac or respiratory failure, even days in advance ([08:55]).

• Data Mining for Research: AI uncovers previously unknown correlations in population health data, transforming fields like psychiatry ([10:30]).
• Drug Development: AI accelerates both novel drug discovery and repurposing existing drugs for new applications by revealing new patterns in large datasets ([11:19]).

Quote:
"L’IA peut proposer des choses qu’on ne connaissait pas avant."
— Giovanni Briganti ([10:30])

3. Opportunities and Ethical Tensions

Interview with Nathalie Grandjean, Researcher at UCLouvain (Medresist project)

• The “4P” Medicine Revolution:
  • Personnalisée: Treatments tailored to individuals.
  • Prédictive: Anticipate diseases in populations.
  • Participative: Patients co-create their healthcare journey.
  • Préventive: Early detection and preventive actions ([13:20]).
• Challenges of Integration:
  • Many hospitals lack the IT infrastructure to integrate advanced AI ("définancements", outdated systems).
  • Tension between AI’s data-driven findings (correlations) and the traditional "Evidence-Based Medicine" (reproducible causal proof) ([15:02]–[16:13]).

Quote:
"L’IA montre des choses qui sont immontrables… si on s’en tient simplement à l’évidence-based medicine."
— Nathalie Grandjean ([16:13])

• Epistemological Friction: AI’s ability to reveal unknown correlations challenges established scientific and medical validation practices, requiring interdisciplinary cooperation ([16:05]–[17:48]).

4. Institutional, Political & Societal Questions

• Need for Strategic Vision: Deployment of AI in healthcare demands a shared political, institutional, and societal vision. Currently, the focus is on competition between hospitals and a general lack of coordination ([18:42]).
• Stakeholders: Effective AI integration needs input from professionals, academia, government, industry, and—crucially—patients/citizens ([18:42]).
• Competency Gap: Despite Belgium’s lead in embedding AI in medical education, lack of competencies regarding AI in clinical workflows is still a barrier ([18:42]).

Quote:
"L’IA est un enjeu de compétitivité... un enjeu de compétence... un enjeu sociétal."
— Giovanni Briganti ([18:42])

• Societal Deliberation Lag: Society’s ability to reflect and regulate lags far behind technology’s rapid advance ([22:39]).

Quote:
"Le temps des technologies n’est pas le temps de la délibération."
— Nathalie Grandjean ([22:39])

5. Education, Transparency & Validation

• Digital Literacy as a Societal Imperative: Calls for comprehensive digital and AI literacy, from early childhood to professional training ([25:33]).
• Transparency: Clinicians must understand the limitations of AI systems to maintain vigilance, avoid overtrust, and ensure better outcomes ([25:59]).
• Validation Crisis: Many AI solutions lack robust scientific validation and are rolled out without the clinical rigor demanded of medications ([26:28]).

Quote:
"La validation des solutions d’IA à l’heure actuelle n’est pas suffisamment bonne d’un point de vue scientifique."
— Giovanni Briganti ([26:28])

6. Additional Ethical and Environmental Concerns

• Cybersecurity: With AI dependent on large, sensitive datasets, hospitals face escalating cyberattack risks ([28:10]).
• Bias and Responsibility: Urges strict vigilance around bias in datasets, transparent algorithms, and dual responsibility for developers and deployers ([28:10]).
• Sustainability: AI’s energy and resource demands (data centers, rare materials) raise major environmental questions ([29:31]).

Quote:
"L’IA, ça consomme des ressources, c’est démentiel… cette course à l’IA... c’est aussi une course un peu aveugle à la consommation d’énergie."
— Nathalie Grandjean ([29:31])

Memorable Quotes & Moments

• On Human-Machine Synergy:
  "Probablement que la combinaison des deux amène à la meilleure prise en charge."
  — Dr. Brouillère ([06:10])

• On the Revolutionary Potential:
  "C’est réellement une technologie qui touche l’ensemble des champs de la médecine."
  — Giovanni Briganti ([07:13])

• On Education:
  "Il faut une alphabétisation au numérique depuis la maternelle… une éducation numérique à l’IA."
  — Nathalie Grandjean ([25:33])

• On Societal Readiness:
  "Le temps des technologies n’est pas le temps de la délibération… ces temporalités sont toujours bouleversées."
  — Nathalie Grandjean ([22:39])

Timestamps for Key Segments

• [00:00]–[06:36]: Radiology field report, limits and benefits of current AI applications (Dr. Brouillère)
• [07:13]–[12:34]: Giovanni Briganti on AI’s pervasive impact in all fields of medicine and research
• [13:20]–[17:48]: Nathalie Grandjean on opportunities and challenges, “4P” medicine, epistemic tensions
• [18:42]–[22:01]: Strategic, institutional challenges and Belgium’s AI leadership (Giovanni Briganti)
• [22:39]–[24:47]: Societal deliberation lag and philosophical reflections (Nathalie Grandjean)
• [25:33]–[26:28]: Education and literacy as critical for adoption and safe practice
• [26:28]–[29:31]: Validation, bias, transparency, cybersecurity and sustainability

Conclusion

AI is quickly becoming a pillar of contemporary medicine—enabling early diagnostics, data-driven research, and individualized care. Yet, the path forward is fraught with technical, ethical, and societal hurdles: integrating AI into legacy systems, harmonizing it with current scientific standards, addressing the digital divide, confronting issues of trust and transparency, and reckoning with its environmental impact. As all guests argue, thoughtful strategy, comprehensive education, robust validation, and inclusive public debate are essential to realizing the promise of AI in healthcare—without losing sight of human fragility and oversight.