Unleashing the Power of Digital Twin Modeling in Precision Medicine Development

Kailash Yagnik, Senior Vice President-Strategy, Siemens Healthcare Private Limited in an interaction with Industry Outlook, shares his views on precision medicine development. During the conversation, he throws light on the issues pertaining to complexity with variations, modification of workflows, and more.

There is a lot of buzz about ‘Precision Medicine.’ What are the factors behind this buzz?

Yes, ‘Precision Medicine’ is quite in vogue nowadays. To understand the reasons, we can first look at what precision medicine means for the stakeholders.  There are four major stakeholders in healthcare: the care provider, the payor, the government and central to these three is the fourth one – the care seeker or the patient.

To the care providers – focus on precision medicine gives ability to deliver care with higher quality, standardize care delivery process, reduce its costs, and increase its reputation in the community.  For the  payor (like insurance, etc.), supporting precision medicine results in overall reduction in cost of servicing the insured, to government, precision medicine would mean lesser loss of productive days of its population, and from a population health management perspective – better prediction and management of epidemics.  

Lastly, and most importantly, for the patient, precision medicine means getting precision diagnosis and a precision treatment which is individualized to his unique needs and characteristics and to his clinical issue.  It also means getting this effective treatment with least inconvenience like duration of hospital stays, or multiple visits to hospital or side effects of medicines & procedures, or overall cost of care, or all of these!

On one hand, there is the objective of precision medicine, while on other hand there are inherently so many variations between hospitals and clinicians on treating the patients.  Is this challenge being understood & managed?

The variations in the processes do pose a challenge in delivering precision medicine.  Here we will need to appreciate that some variations are inherent to the type of patient disease, anatomy & physiology, and hence, we should be referring to “Unwarranted Variations” that arise, essentially due to ways in which “man” and the “machine” impact the delivery of care.  Also, managing these unwarranted variations should not be seen as separate topics,  rather as an integral pre-requisite for delivering precision medicine.

These unwarranted variations stem from the specific training and habits of individual physicians and technicians, while others reside within hospital technology and equipment. The net result is that these variations increase healthcare costs, which could be up to 25 percent, and decrease quality of care.

Managing unwarranted variations requires a comprehensive, unbiased, and enterprise-wide approach.  A few of the topics that care providers typically consider are:

o  They accept that reducing variations early in the steps of care process greatly reduces the cumulative impact of these variations.

o They adopt local or global evidence-based best practice standards in delivering care.

o They prefer technologies that can adapt to patients’ individuality and deliver consistent results for all patient types. Example: patients who are obese, geriatric, immobile, physically unstable, etc.

o They deploy automation, as much as possible, to reduce operator related variations.  Example: auto set-up of CT and MR exams, however ultrasound exams still remain operator dependent! 

o Care providers have also started considering the use of AI based algorithms for “Assisted Decision Making” in order to reduce clinical variations.

Consider the following example as a case in point on how automation technology can assist clinicians.  For cardiovascular artery diseases, Percutaneous Coronary Intervention (PCI), or also known as Angioplasty, is the gold standard for treatment.  In this procedure, cardiologists use a catheter to insert one or more stents in order to open narrowed coronary arteries. Historically, PCIs have been performed manually, a feat of precision and expertise that is a testament to the skill of the specialists who perform them. However, significant variability exists in skills of the specialists. Some are simply more experienced and skilled than others, and this fact could influence the quality of outcome.

However, a Precision Medicine tool - Robotic-Assisted PCI (R-PCI) can combine the advantage of human judgment, intuition, and decision-making of the cardiologists with precision, control, repeatability, and procedural automation of a robot. The movements of the catheter and the stent are controlled by the interventional cardiologist, but they are executed by a robot that does not get tired, does not get uncomfortable, does not feel stress, and does not get distracted. Hence, specialists can deliver consistent and high-quality treatments, case after case and day after day.

You talked about some variations that are inherent to the complex biological process of the disease.   Are there any efforts to tame these variations as well?

Indeed, the biological process behind diseases are complex and to complicate the matter further, for the same disease this complex process also varies from one patient type to other. This creates an interesting challenge for the goal of precision medicine. Hidden in this challenge is the opportunity – that if we  can successfully deconstruct these variations in the biological process, then both the diagnosis and the treatment can be effectively tailored to the needs of the individual.

One way to do this is by using genetic testing to identify the specific causes of a person's medical condition. This can help healthcare providers select the most appropriate and effective treatments for that individual, rather than relying on a one-size-fits-all approach.

The other big enabler in understanding the disease complexity is Artificial Intelligence (AI).  If precision medicine is the destination, then AI is the road to it.  As we speak, already many AI algorithms are being developed and are constantly learning from the huge repository of patient data that is available and that which is  continuously getting generated every day.

Delivering ‘Precision Diagnostics’ will soon be a multi-modality approach. Big data and use of AI algorithms will permit combining data from variety of pathology tests, genomic tests, medical imaging, radiomics, proteomics, wearable devices, etc. and integrate it with individual differences in lifestyle and environment, to give an unprecedented view of the disease process as it stands on that day.  Further, matching this view with the repository actual case development of same patient type, will allow us to see the evolution of disease process making a ‘Precision Prognosis’ also possible.

The data being collected today will be the foundation to improving precision medicine of tomorrow, and without the right use  of AI, factoring complex disease process for effective delivery of precision medicine cannot be done.

To deliver precision medicine, do the care providers need to modify their existing workflows?

A short answer is, yes.  Realizing the promises of precision medicine will need a relook at the existing workflows, first from the perspective of reducing the variations,  adopting right technology and AI.  It will also need revisiting of all the process steps in between – commonly known as “door-to-needle” or better understood as patients’ entry to treatment.

Let us see one more case, this time in Oncology.  Cancer is a disease that presents unique challenges for healthcare providers because of its wide variety and its complexity. The goal in effective cancer care is to have shortest time possible between diagnosis and treatment of the disease.  As an example, in early-stage breast, lung, renal, pancreatic cancers, a four week of delay in treatment, increases the risk of death by approximately 10 percent.  Minimizing these delays could improve cancer survival rates.

Radiation therapy is one of the commonly used treatment methods for cancer. Various factors in the hospital system, like overburdened healthcare systems, shortage of medical staff, lack of coordination between different departments treating cancer, etc. cause delays in the diagnosis, planning and patient treatment. In many centers, time from consultation with experts to the first day to getting the treatment, can often run into weeks.  However, a significant reduction in this delay can be achieved by adopting AI and recasting the workflow.

It has been demonstrated at Amsterdam UMC Cancer Center where, a team under Dr. Wilko Verbakel, MD, a senior medical physicist, was able to “fast-lane” the treatment time for FAST METS type of cancers, from weeks to just a few hours.  They have achieved this by doing away with the step of “CT Imaging for Simulation”, by using advanced imaging technology right on the treatment machine, and by pre-planning the treatment using AI based automated identification and outlining of organs-at-risk, such that only minimal additional efforts were needed for actual treatment.  The combination of these factors also enabled the treatment to be adapted daily to the actual condition on the patient on that day, and thereby enabling precision medicine.

How do you see Precision Medicine evolving in the near future?

Going forward, a few of the developments that we can expect are:

Understanding organ-specific risk based on genomic data and comparing it with real-world data collected from previous patients could drive us towards a more agile, data-driven, and impactful precision medicine.

It could be possible to extend the FAST METS, the Amsterdam example that we previously saw to other cancer patient groups where faster access to treatment will directly improve their survival chances.

Due to early detection, the surgical procedures will be needed to remove much smaller tumors. Specialization, greater expertise and precision will be key along with superior control and navigation of the surgical tools. Likewise, interventional procedures, such as PCI that we saw earlier, will continue to require a high degree of precision and accuracy, navigating complex, tortuous blood vessels, and properly sizing and positioning stents on the first attempt. This means, robotics will be more entrenched in these specialties. They will not replace humans but will augment human capabilities.

Let us now see the most interesting facet of precision medicine. It will become possible, in routine clinical practice, to combine real-time anatomical, pathology data of a patient’s organ in order to create its digital model. To this, it would be possible to add the physiological information (like flow, etc.), thereby getting a Digital Twin of that organ. A fully developed patient twins will benefit clinical decision making for diagnostics as well as treatment. This digital twin will allow clinicians to safely plan lifesaving, drug treatments, surgeries, implants, etc. and pre-ascertain their efficacy before implementing them on the patient. This very “unique to patient” digital twin modeling will take precision medicine to a next level.