From scientific reflection to concrete planning. This is the paradigm shift that emerged from the workshop. "AI & Forensic Medicine", hosted by theUniversità Campus Bio-Medico di Roma on the eve of 47th National Congress of the Italian Society of Legal Medicine and Insurance (SIMLA), scheduled from June 18 to 20 at the Conference Center of the Catholic University of the Sacred Heart.
The initiative brought together the medical-legal community, artificial intelligence experts, and industry representatives with the aim of launching a collaborative effort to develop technological tools that are truly useful for medical-legal practice.
The professor opened the proceedings. Vittoradolfo Tambone, Professor of Forensic Medicine and Bioethics of theUniversità Campus Bio-Medico di Roma, which highlighted the concrete approach of the initiative: "The workshop had a pragmatic goal: not to discuss the big issues of artificial intelligence, but to understand how to work better in the field of forensic medicine using AI, not only as end users but also at the algorithm design level." Precisely for this reason, the workshop also involved IBM Consulting, with the aim of launching a shared development path for future artificial intelligence applications. "One of the workshop's conclusions was precisely to engage, together with a technology player, in the co-design of algorithms."
Secondo Tambone, forensics must take an active role in the development of artificial intelligence."Don't stop the wave, nor even suffer it: learn to ride it. What we provide is a surfboard, so that forensic pathologists, those involved in risk management, bioethics, and forensic psychiatry can use it to the fullest." And yet: The key shift we'd like to see is from the "human in the loop" paradigm to the "AI in the human loop." This requires us to address the system at the purpose and design level, so as to build a tool that works for science, for scientific truth, for justice, and for the common good. This way, we'll ensure that the tool is built specifically for science, not science for the tool.
The workshop thus outlined a new methodological perspective: Artificial intelligence systems intended for forensic medicine will have to be born from a structured collaboration between forensic doctors, data scientists, engineers, jurists and technology companies, So that the discipline's expertise contributes from the algorithm design stage. In this vision, AI is not a shortcut that replaces professionals, but rather a "reinforcing alteration" that expands their analytical, reconstructing, and interpretive capabilities, according to a new "hybrid forensic methodology." The guiding principle is the ethic of work well done, founded on strengthening critical thinking.
The strategic value of the initiative was also discussed by president of SIMLA, Francesco Introna, who introduced the workshop as "a fireworks display to open the congress", announcing "a working group on artificial intelligence branded SIMLA to begin to indicate to programmers which are the reference parameters to use in the various fields". The President also emphasized that Congress is looking to the future prospects of the discipline. Unlike in the past, this is a congress projected toward the future: we're not talking about achieved realities, but about the future projection of forensic medicine. And one of the key themes of the near future is certainly the approach of artificial intelligence, where we don't have major diagnostic problems, but rather problems in the projection of our diagnoses and our activities, with obvious judicial and evaluative implications, and with errors that can arise from an incorrect application of AI.
Among the main critical issues that emerged is the reliability of digital evidence in judicial proceedings.The coroner presented with an x-ray, remade from scratch with artificial intelligence, is unable to unmask it", explains Introna. "And since the magistrate relies on what the coroner says, we would risk unjust convictions or improper acquittals.". The solution, according to the SIMLA president, involves controlling what enters the systems: "We need to look inside the black box, because only by inserting adequate material will we receive adequate answers, otherwise, the outputs will inevitably be unreliable. ". Introna clarifies the border: "It's fine to use AI to report historical and clinical data, but from the physical examination onward, and especially in the evaluation, the decision is the product of the expert's processing. They can use artificial intelligence to broaden the picture, but the final decision, the one that must stand up in court, is theirs and theirs alone. Otherwise, it's an ideological forgery committed by a public official.".
A look at the technological prospects was offered by Luca Di Piramo, Senior AI Specialist at IBM Consulting. "The evolution of AI is changing the relationship between people, processes and data - observe -. Traditional AI requires comprehensive data to train specific models; with GenAI, the focus shifts to the quality and certification of the information context that generalist models still rely on. Agentic AI goes further: it doesn't just understand, but coordinates processes and interacts with external systems, requiring interoperability and more robust flow management.". A reflection that reverberates in the healthcare sector in a specific way: " In this sector, the role of the professional is transforming, becoming an active supervisor and clinical guarantor. Guardrails, traceability, and human oversight of the entire process become essential. If GenAI amplifies risks to content, agentic AI amplifies those to actions.". It is a decisive change because it passes "from systems that predict, to systems that understand, to systems that act and governance must become increasingly end-to-end to safely accompany this new operational capability".
SIMLA National Secretary Lucio Di Mauro He outlined the next steps in the process. "The next step will be to transform scientific discussion into concrete planning: stable working groups, lines of applied research, dedicated training programs, and shared operational tools." Innovation, including generative artificial intelligence, warns Di Mauro, must be governed with moderation because "It must enter forensic medicine not as a shortcut, but as an opportunity to raise the quality of medical-legal services, in compliance with its essential ethical and deontological principles, preserving freedom of scientific expression, always based on evidence and oriented towards offering a contribution to the search for truth in the interest of the entire community.".
The President of the Congress Paola Frati he recalled the value of the workshop as a moment of discussion dedicated to one of the main challenges of the discipline. "It was a highly sought-after pre-conference event entirely dedicated to a highly topical topic: artificial intelligence. Thanks to the workshop's scientific director, Professor Tambone, we were able to engage with a key player like IBM to understand how forensic medicine can concretely utilize AI without being subjected to it, creating programs that can develop it in a practical way. We discussed its application in emerging fields: not only forensic pathology, but also, for example, clinical risk and risk management."
The works were closed by Anthony Olive, President of the Congress and full professor of Forensic Medicine at the Catholic University of the Sacred Heart, which drew attention to the delicate balance between technological innovation and procedural guarantees. "In the criminal sector, where scientific evidence directly impacts people's freedom, forensic medicine must be able to define which biological or forensic phenomena require advanced analysis, which variables should be considered relevant, and which margins of uncertainty are compatible with the principle of reasonable doubt.". Only in this way, he underlines Professor Antonio Oliva, artificial intelligence "It can be useful if it works with traceable, verifiable, and reproducible data, in full compliance with due process guarantees. Forensic medicine, in fact, deals with unrepeatable events and complex contexts: the specialist's experience allows for the integration of data, uncertainties, and legal implications. The goal is not to delegate, but to enhance the quality of the investigation.".
