True Human Digital Twins are a unique technology that uses clinical, biological, mental, environmental, and social data to create a unique, real-time digital representation of an individual’s entire health profile. Using AI and machine learning, they act as an early warning system for lifethreatening illnesses, detecting emerging risks 72 hours earlier than traditional diagnostics.
1. How do you define a “true human digital twin” in contrast to traditional digital health models or digital patient records, and what makes it fundamentally transformative for healthcare delivery?
A true human digital twin (HDT) comprises three crucial components: access to an individual’s personal health data, a certified AI model that uses this data to predict specific outcomes, and a mechanism to deliver these results to clinical teams for enhanced patient care. This approach builds on traditional healthcare delivery by providing powerful, actionable insights to empower overstretched clinical teams, giving them the relevant patient information, emerging risk predictions, and recommendations for the best actions to take.
The ever-increasing challenges facing healthcare systems today, significant healthcare professional shortages, coupled with an ageing population with more complex ailments, mean that relying on established care models can no longer ensure the high standard of care patients expect and deserve. The transformative aspect of an HDT lies in its ability to support clinical teams to deliver higher quality, person-centered care without drastically altering their existing workflow, serving as Clinical Decision Support.
The HDT approach is distinct from other digital models, as it provides a full 360 overview of the patient and models out the appropriate actions based on the totality of evidence available etc.
2. The concept involves integrating clinical, biological, mental, environmental, and social data into a single digital construct. What are the biggest challenges in harmonising such diverse and complex datasets into a reliable and real-time health representation?
The main challenges we have found are:
- Accessing, linking and standardising patient data: in addition to various levels of data-sharing across healthcare practices, vast amounts of our personal health data reside in siloed databases and are saved in various formats, creating a complex and labour-intensive process to create unified datasets.
- Building AI models that are fit-for-purpose and that can interpret the temporal value of data. For example, models that can assess the comparable relevance of blood test results from two years ago to current measures available, can provide clinically meaningful recommendations..
- Cybersecurity: of course, throughout this process, robust data protection and cybersecurity measures are crucial, with governance structures in place to ensure the safeguarding of patient privacy.
3. From a technical perspective, what role do AI and machine learning play in ensuring that a human digital twin is both predictive and adaptive to an individual’s changing health profile?
Our approach leverages a variety of futuristic techniques, including continuous and adaptive learning, integrated within our platform, to ensure the HDT operates consistently within regulated boundaries. In addition, we have pre-established robust mechanisms to retrain models automatically, maintaining accuracy and compliance.
4. One of the most striking claims is the ability to detect life-threatening illnesses up to 72 hours earlier than conventional diagnostics. Could you explain how this predictive window is established and validated in real-world clinical settings?
Similar to how we know that symptoms like an itchy throat, fatigue, or loss of appetite can appear a couple of days before developing the flu, HDT can identify early trends that indicate a patient’s changing condition by analysing several clinical data points live and in situ, eliminating the risk that these subtle signals will go unnoticed.
In collaboration with clinical teams, we have defined the ideal “window of opportunity” – the optimal time to alert HCPs to potential risks with clinically relevant sensitivity and specificity, thus prompting timely and effective action. Our models are capable of predicting risks up to seven days in advance. Through clinical co-design, we found the optimal alert window to be 12-36 hours, providing sufficient time to administer the right treatment or prevent ward closures by isolating patients promptly and balancing the risk of false positives, avoiding alert fatigue among clinicians.
Our latest feature, soon to be released, allows clinical teams to customise their preferred alert window and sensitivity. This flexibility enables teams to customise the alert frequency based on the individual patient needs; for example, it may be preferable to have earlier, more sensitive alerts for high-risk patients on respiratory wards, while general ward teams may prioritise higher accuracy.
5. Real-time monitoring at such a holistic scale raises questions about interoperability with existing healthcare infrastructure. How do you envision seamless integration of digital twins into current hospital systems, electronic health records, and diagnostic workflows?
We are able to integrate the HDT directly into clinical workflows by partnering with several leading Electronic Patient Record (EPR) providers to securely embed MEMORI and create a seamless experience for clinical teams who can continue using their familiar tools without disruption.
6. What ethical considerations come into play when creating a digital representation of an individual’s entire health profile, particularly regarding data ownership, consent, and patient autonomy?
Gaining consent to share patient data is a highly complex and often difficult process, but at Sanome, we’re acutely aware of the sensitivity surrounding accessing and handling such personal data, and don’t take this responsibility lightly.
Over 82% of patients support the use of their data for clinical research and the development of better tools to improve care, provided that appropriate governance structures are in place.
To ensure data is shared safely and ethically, we have worked closely with several Ethics Committees, Confidentiality Advisory Groups, and patient advocacy groups to establish robust governance and oversight frameworks to ensure that data is shared safely and ethically:
- Data ownership always remains with the individual, while the healthcare provider (in this case, the NHS) acts as the Data Controller.
- We have implemented strict data de-identification protocols to protect individuals.
- We acknowledge and uphold patients’ rights to opt out at any time.
7. Environmental and social determinants of health are often overlooked in medical diagnostics. How does incorporating these dimensions into a human digital twin improve early detection and overall patient outcomes?
Absolutely, these factors play a significant role in patient health. For example, high pollen levels can worsen conditions like asthma or chronic obstructive pulmonary disease (COPD), increasing the likelihood of hospital admissions or the need for additional oxygen treatment. Similarly, elevated temperatures can contribute to blood pressure complications, raising the risk of cardiovascular issues.
Additionally, unfortunately, lower socioeconomic status is often linked to poorer health due to a range of factors such as inadequate housing, underlying health problems, and poor nutrition.
Our HDT model takes these additional factors into consideration, along with the patients’ personal and clinical data (health profile, mediEXPERT TALK www.europeanhhm.com 77 cal records and any wearable data) to provide healthcare professionals with a full picture of the patients’ overall health, as well as accurate insights and actionable recommendations.
8. Predictive models in healthcare are often criticised for a lack of explainability. How does your system ensure transparency so that clinicians can understand, trust, and act upon the insights generated by a human digital twin?
From day one, we have prioritised transparency and clear communication. MEMORI clearly highlights the key data points that explain why the model predicts a certain outcome – we call this the AI explainability. These have been co-designed with clinical teams to ensure alignment with real-world applicability and to enhance readability and interpretability.
Furthermore, all of our AI models undergo rigorous evaluation, auditing, and certification by our Notified Body, and our results are submitted for peer review to ensure suitable external scrutiny and best practice.
9. Beyond early detection, how do you see human digital twins evolving in terms of preventive care, personalised medicine, and long-term disease management?
The future we envision is one where everyone has access to their own HDT that serves as both an entry point and triage tool in healthcare, easing the strains on healthcare services and suggesting personalised recommendations for the best course of action for the individual based on the information gathered, all under the oversight of clinical teams across community care, pharmacies, general practice, and secondary care. This could include personalised recommendations for preventative care and long-term disease management, such as prescription reminders based on real-time and real-world insights into current health conditions and environmental factors.
10. In terms of scalability, what infrastructure and policy frameworks are required to bring human digital twins from proof-of-concept to widespread adoption across diverse healthcare systems worldwide?
There are several key factors to consider. First and foremost, MEMORI is classified as a Software-as-a-Medical Device and is therefore highly regulated by health authorities such as the Medicines and Healthcare products Regulatory Agency (MHRA) in the UK, the Food and Drug Administration (FDA) in the US, and the EU Medical Device Regulation (EU: MDR) in Europe. As of April 2025, MEMORI is a CE-marked Class IIb medical device; the first real-time infection-prediction tool cleared at this level in Europe, while maintaining compliance with section 251 regulations on the use of patient data. These AI advancements are more than futuristic ideas, and we take our responsibility to patients seriously.
As our technology is embedded within existing EPR providers and leveraging their established infrastructure, we can ensure seamless integration and scalability for clinicians and the health system.
In order to bring this advancement to clinicians and patients in the real world, it needs to be sustainable. Our reimbursement model is focused on ensuring our platform provides actual value to clinicians and patients, and we operate on a shared benefit agreement approach.
11. Could you share examples or case studies where human digital twins have already demonstrated significant impact in clinical decision-making, patient outcomes, or healthcare efficiency?
So far, there are no large-scale case studies of complete HDTs in daily clinical practice. However, adjacent systems like multimodal risk stratification platforms, early warning systems, and precision prescribing platforms are some examples where AI is having a significant impact on earlier disease detection, reduced clinical workload, and better decision-making.
12. With constant advancements in genomic medicine, wearable technology, and real-world evidence, how do you see these fields converging with human digital twins to create a more precise and individualised healthcare ecosystem?
These are fantastic developments, and we are extremely excited about the opportunities they present. The key to success will be integrating these advancements into our HDT, enabling seamless support for clinicians’ decision-making at the point of care, directly within the EPR. Ultimately, this will empower better, earlier, and more personalised clinical decisions.
13. Looking ahead, what do you consider the greatest opportunity – and the greatest barrier – in ensuring that true human digital twins become a standard of care rather than a futuristic concept?
We are already delivering on this vision, delivering the future of healthcare today, with several partnerships with the NHS and private hospitals.
The greatest challenges and opportunities go hand-in-hand; initiatives that promote better and easier data sharing, enhance interoperability between EPR vendors, and address the significant cost pressures facing healthcare to deliver improved care more efficiently will provide a more seamless overall experience.
MEMORI has the potential to support 55,000 clinical decisions across millions of unique patients. Our goal is to continue building on the work we have initiated with our pilot programmes; delivering bespoke clinical decision support and providing personalised recommendations to clinicians, easing the burden on healthcare services and empowering the system to shift from a reactive to a proactive model of care.