Our team member Henrik Podéus Derelöv, together with his colleagues at Linköping University and the Swedish National Board of Forensic Medicine has developed a digital twin model designed to improve the forensic reconstruction of alcohol consumption. By integrating individual physiological characteristics—such as sex, age, height, weight, and medical conditions—with data from breath, blood, and urine samples, the model enables more precise estimations of when and how much alcohol a person consumed. This approach paves the way for more reliable investigations into crimes where alcohol is believed to have been involved, and it addresses limitations in current forensic methods, particularly in cases involving the “hip flask defence,” where individuals claim alcohol was consumed only after the incident. The model also incorporates factors influencing alcohol absorption and metabolism, including gastric emptying, food intake and the type of alcoholic beverage consumed. Although the framework does not eliminate uncertainty, it offers a more flexible and reliable tool to support forensic investigations involving alcohol-related crimes and accidents.
Henrik Podéus – PhD student at the Department of Biomedical Engineering (IMT) and ISB Group member behind the research
Gunnar Cedersund held a presentation on our digital twins at the annual event “Healthy Vascular Aging Symposium”, at Danderyd’s hospital, on April 21. The event was arranged by the Swedish Organization for hypertension, stroke, and vascular medicine, which is a national network for scientists and clinicians. Apart from this annual event, they also arrange a study group for Ph.D. students in the field.
This year’s event had a special focus, and apart from Gunnar’s presentation on digital twins and Henrik Hellqvist’s presentation on machine learning (photo below), there was an interesting lecture by Ljubica Matic, from Ulf Hedin’s group. Together, these three presentations created a nice flow: the large database for vascular health built up by the Hedin and Matic, traditional ML-based analysis of such data, and our own hybrid approaches, including all other organs and longitudinal data, among others.
At the event, we also went through our upstarting and exciting M4-HEALTH project, which includes digital twins not only for vascular health, but also for all other medical applications, including even non-clinical actors. Read more about our excellence cluster on m4-health.se and don’t miss our next workshop, on May 25-26.
Today, April 29, Gustav Magnusson defended his Ph.D. thesis Cerebrovascular reactivity measurements in patients with Aneurysmal Subarachnoid Hemorrhage. Gustav has been co-supervised by Gunnar Cedersund and Maria Engström, who have collaborated for many years on modelling of the brain, with a focus on modelling and understanding the BOLD and fMRI-signals, using multi-modal MRI data.
In this Ph.D. project, we have expanded the MRI-side of this collaboration, to create a new measurement approach to monitor subarachnoid hemorrhage patients. The underlying problem is that these patients, who have gotten a severe type of a stroke, have a high risk of getting a second stroke, around 7 days after the initial bleeding. Therefore, these patients are in intensive monitoring at the neurointensive care unit.
In this project, we have tested a new approach to monitoring: to combine CO2-perturbations with measurements of blood flow. The basic hypothesis is that patients that are on the verge of having a secondary ischemic event (a second stroke), show a lower response to CO2-perturbation, indicating stiffer and less responsive blood vessels. Gustav has developed this technology, and tested it on both healthy volunteers and real patients. As a next step, we are now going to start modelling of these data, to understand the mechanistic basis for these processes.
This project was funded by the Brain Foundation and by the EU-project STRATIF-AI, which focuses on stroke, and which is coordinated by Gunnar Cedersund. Maria Engström was the main supervisor on paper. However, the main supervisor in practice was Anders Tisell, an MR-physicist at Region Östergötland. The two other co-supervisors were the neurosurgeon Lovisa Tobieson and the neurologist Charalampos Georgiopoulos.
Congratulations Gustav to a successfully defended thesis!
Understanding when—and how much—alcohol has been consumed is essential in both healthcare and forensic contexts. Yet this remains a surprisingly difficult task. Self-reported drinking data can be unreliable due to stigma, impaired memory, or legal consequences. Objective biomarkers such as blood- and breath alcohol concentrations (BAC and BrAC) are widely used, but they decline rapidly, making it difficult to reconstruct past drinking events with precision. At the same time, slower biomarkers like ethyl glucuronide (EtG), ethyl sulphate (EtS), and urine alcohol concentration (UAC) provide valuable complementary information, but existing approaches rarely combine these data sources in a meaningful, mechanistic way. In our newly accepted article, “A Digital Twin Framework for Forensic Reconstruction of Alcohol Intake via Fast and Slow Metabolite Kinetics,” we present a solution to this long-standing challenge.
We introduce a physiological digital twin that integrates the kinetics of both fast-acting and slow-acting alcohol biomarkers into a single mechanistic framework. This allows us to: • Capture the joint dynamics of BAC, BrAC, EtG, EtS, and UAC • Reconstruct past drinking events with improved temporal precision • Support forensic evaluations, including scenarios like refuting the hipflask defence • Enable personalized simulations of alcohol metabolism based on individual physiology By linking these biomarkers mechanistically rather than statistically, the model leverages the strengths of each marker, improving inference in situations where forensic accuracy is critical.
Improving Forensic Reconstruction
Our digital twin demonstrates clear benefits: • Higher reconstruction accuracy: Combining markers reduces ambiguity in the timing and quantity of consumed alcohol. • Robustness for complex scenarios: The model supports analyses requiring fine-grained temporal resolution—something single-marker models struggle with. • Practical usability: We provide an interactive web tool where users can explore hypothetical drinking scenarios and visualize expected biomarker trajectories. This tool is designed for forensic analysts, clinicians, toxicologists, and researchers interested in alcohol metabolism and its quantitative reconstruction.
Bridging Clinical, Forensic, and Computational Domains
We see this work as a foundation for precision modelling in alcohol kinetics. By moving toward unified digital twins grounded in physiology, we can support decision-making in healthcare and legal settings with more transparency, reproducibility, and individualized insight. We’re excited to share this contribution with the community and look forward to continued collaborations around modelling, biomarker research, and forensic applications.
If you’d like to try the interactive simulation tool, feel free to reach out!
Is knowledge enough to create better life conditions for everyone? I remember the PhD proposal I wrote three years ago: I stated in that proposal that I did not want to do science just for the sake of knowledge. I wanted to do science to make better healthcare possible in the real world – here and now – and make it inclusive, centered around an individual. I did not get the PhD position. Instead, life blessed me with this:
On January 28, 2026 I traveled to Linköping University Hospital to join my colleagues – visionaries – not only one of a kind experts in their respective fields but also revolutionists who are building the future of healthcare – the healthcare that each and every one of us, no doubt, has wished for at least once when we have dealt with healthcare systems: A health ecosystem that treats each and every one of us as humans whose unique health journeys are formed through equally uniquely intertwined factors – information that current healthcare system does not have the capacity to gather, not to mention the capacity to help individuals or healthcare professionals use it as actionable data for disease prevention.
I was there as the Press Officer of the world-unique Medical Digital Twin which allows individuals to gather their physiological and cognitive health data – from the intracellular level to the whole-body response – to account for all major functionalities in the body, make informed lifestyle decisions now and predict how one’s health state will develop over years.
So why did we gather in January? With Medical Digital Twin at its heart, we have built M4-HEALTH – an excellence cluster bringing together an even larger pool of experts in healthcare, drug development, defense, forensics, elderly care, AI and data handling as well as food and textile industries to transform our current reactive and hospital-centric healthcare system into a preventive, personalized, ecosystem-based approach.
To go back to my rhetorical question at the beginning – Is knowledge enough to create better life conditions for everyone? It takes more than knowledge. In M4-HEALTH we have world-class knowledge and expertise. We have hearts that beat for change. And we have a vision that we are all fiercely dedicated to, no matter what role we are currently playing in the excellence cluster. I believe that is that foundation of using knowledge for the best for everyone.
And if you want one more hint of a personal touch… When I was a little girl and breathed books like one breathes air, I always envisioned myself doing revolutionary things. But I had no idea this is where my life would take me.
Speaking of visionaries, here are two people that inspire me: Gunnar Cedersund that has united so many of us around his vision of what the future of health can look like, and Johan Holmsäter that has revolutionized our perspective on physical health once and is on his second round.
Tomorrow Thursday, Gunnar will present an overview of our digital twins for the society SeniorNet, which deals with IT-related issues for seniors (60+). This lecture is a part of our growing initiatives towards creating a health-centric ecosystem for elders. This ecosystem will involve both these types of societies and member organizations for elders and their interests, together with traditional healthcare organizations (e.g. primary healthcare clinics, geriatric clinics, cardiology and internal medicine clinics, etc), informal healthcare actors (gyms, health coaches, yoga teachers, etc), eldercare homes (like our partner organization Norlandia), assisted living organizations, municipalities, regions, etc. At this time, I will give a glance of this growing ecosystem, which also involves our second spin-off company – Lev Skönare. I will also outline how our digital twins can help to facilitate the integration between all of these actors. All of this is a part of our big excellence cluster M4-HEALTH, which will create a new “Model for Health”, centered around our digital twins, not only for these actors related to 60+ members of society, but for all ages and actors related to health (more about this in future blog posts). I will mention also this new and very exciting initiative at this lecture. Finally, I will also mention our upstarting collaboration with the very interesting Måltidsgillet, which is a very successful initiative from Skåne, dedicated to tasty, healthy, and affordable 3 course lunches for elders, also involving lectures, healthy activities, etc. Welcome! 🙂
Program for tomorrow, taken from their home page. Link here. As you see, there is also a lecture by Åke Hultgren on smart homes, with integrated sensors, etc. This item actually also relates to projects we have on elders, and on treatment for heart failure patients in Australia.
On September 1–2, our team attended the Strategic Research Area in Forensic Sciences (SoFo) Annual Meeting 2025, held at the scenic Rimforsa Strand. This lunch-to-lunch event brought together researchers, professionals, and collaborators from across Sweden to share insights, foster new connections, and explore the future of forensic science.
Presentations from our group
This year, our group contributed with two presentations highlighting ongoing work within forensic toxicology and post-mortem analysis:
William Lövfors, with support from Henrik Podéus, presented the alcohol model developed to aid in estimating alcohol intake in forensic investigations. The model represents a valuable step toward improving the accuracy and reliability of alcohol-related case assessments.
Rasmus Magnusson shared his research on estimating time since death using post-mortem interval (PMI) data and machine learning models – a project currently under review in Nature Communications. His talk sparked great interest and discussion around the application of advanced computational methods in forensic medicine.
Collaboration and future directions
Beyond the formal sessions, the meeting offered many opportunities for discussion, idea exchange, and networking. Participants engaged in lively conversations around potential research overlaps, interdisciplinary approaches, and strategies for expanding SoFo’s scope in the coming years.
The meeting reaffirmed the strong sense of community within SoFo and underscored the exciting directions in which forensic science is moving—through collaboration, innovation, and shared scientific curiosity.
We thank the organizers for an inspiring and well-arranged event, and we look forward to continuing the conversations and collaborations sparked at Rimforsa Strand.
Maria-Anna Sotiropoulou held a session on digital twins, explaining their relevance in various fields, including healthcare, at the “Girls in Biomedical Engineering” workshop on 19th of November at LiU Campus Valla. The aim of the workshop was to inspire and foster interest among young women aged 17-18 in engineering, technology and their application in real-world scenarios, particularly in health and wellness.
After the presentation, the girls participated in a hands-on activity where they measured their blood pressure using digital sphygmomanometers (automatic blood pressure monitors). They entered these readings, along with their anthropometric data – such as height, weight etc.- into their digital twin model. The model was designed to simulate and analyze potential stroke risks and it also included simulations of weight changes and blood pressure fluctuations, allowing the participants to visualize how these factors could impact their health over time.
Workshop participants entering their blood pressure readings, along with their anthropometric data, into their digital twin models.
