The medical condition of digital twins

Without wishing to pick favourites among our clients’ sectors, healthcare has to be one of the most exciting and meaningful areas for AI deployment. Whether it’s instantly analysing x-rays, accurately diagnosing conditions, speeding drug discovery or assisting actual surgery, AI has the potential to revolutionise patient treatment and institutional operations. 

Of course, it has its risks. Privacy and data quality are key concerns in AI deployment in any organisation. As with any new technology, it would be unwise to let it run without controls and supervision. That’s why technologists are developing AI tools to work through virtual scenarios, simulating medical implementations of AI before they involve actual patients.

Long-standing concept with novel applications 

While AI is a relatively new technology, this idea has been used in industry for decades. In the 1960s, aerospace manufacturers created digital twins to work out how their craft might withstand the impact of leaving earth’s atmosphere. Years after the space race, they are a standard part of R&D and factory operations across sectors. Enabling products to be tested virtually determines the best use of materials and labour, predicting faults and necessary maintenance. 

This concept is now being applied in healthcare to ensure AI deployment is appropriate, successful and to avoid harming patients. Using data on hospital capacity, staffing and equipment, researchers from Seoul National University Hospital and Harvard Medical School have designed the world’s first virtual hospital framework. The AI powered virtualisation takes the role of an all-seeing director that can allocate resources, improve workflows and maximise operational efficiency.

Digital twins are also being developed on an individual patient level. Where AI can help diagnose conditions and recommend treatment, a simulation will generate various paths and outcomes based on disease trajectory templates, without risking harm to real patients.

Digital twins have a role in improving representation in medical research

However, it must also consider the bias that exists in areas of research when it comes to race, gender, age, and other factors. For example, researchers at Aix-Marseille University point out that female immune systems respond differently to certain drugs. Despite this, biomedical research has largely been conducted on male biology. 

Medicine is often a generalised practice where treatments are based on average responses observed in large populations. Therefore, digital twins that simulate how a disease might develop and how treatments could work can only be meaningful if they have a complete picture based on an individual’s history and physicality. 

Aix-Marseille’s researchers say that personalised models capturing the unique biological characteristics of each individual are under work across Europe as a way to advance precision medicine. The resulting hyper-personalised care plans based on personalised models for digital twins can only improve patient outcomes. The speed and scalability of AI means that we can hope for a future where it does this across populations.

Moving from theory to implementation

All round, AI in healthcare is a very exciting prospect. It promises not only to take the admin burden from professionals so they can get on with caring and clinical treatment, but also to speed processes, increase accuracy in diagnosis and stop serious symptoms being missed. 

As our non-healthcare clients say, the success of AI models is down to the data they are fed – not just accurate data but complete data. Digital twins can keep a check on the AI tools and systems being deployed in healthcare to make sure they behave themselves. Powered by AI, they could provide personalised care on a mass scale.