Digital twins to predict disease

Vir­tu­al real­it­ies are a way of extend­ing and improv­ing the real world, and this is also the role of a digit­al twin. In tan­gible terms, they provide a frame of ref­er­ence upon which dif­fer­ent dis­cip­lines can col­lab­or­ate on a com­mon pro­ject. By com­bin­ing their know­ledge and expert­ise, they will be able to share, on the basis of the same rep­res­ent­a­tion, sev­er­al pos­sible scenarios.

Nobody expects a digit­al twin to be per­fect. To draw a par­al­lel with the auto­mot­ive industry, 95% of crash tests on cars on the mar­ket are car­ried out on com­puters. Thanks to the digit­al twin of a vehicle, even if it is incom­plete, we can test its res­ist­ance to a crash, for example – hence, these are mod­els that can be trus­ted. The human body is obvi­ously more com­plex than a car, but our object­ive is sim­il­ar: we want to improve life expect­ancy and keep people healthy at a cost that is sus­tain­able for the health care system. 

To do this, at Dassault Sys­tèmes we have imple­men­ted a real flex­ib­il­ity between the dif­fer­ent scales and dis­cip­lines that are used by these vir­tu­al tools. Our digit­al twin of the heart is defined as multi-phys­ic­al. The heart is a pump, and there­fore has mech­an­ic­al beha­viour that make it con­tract when stim­u­lated by an elec­tric­al sig­nal. We can digit­ally rep­lic­ate this phe­nomen­on and the way the pump is con­trolled. As such, this twin is also cap­able of repro­du­cing the geo­metry of the heart from an ima­ging sys­tem, but also the beha­viour of the muscle fibres: this is the first type of phys­ics that we have used.

Although a mod­el is can only make use of the data you put into it, it can still be used to make predictions.

This mod­el heart con­tracts, cir­cu­lat­ing the blood. From there, we went on to call upon oth­er types of phys­ics to repro­duce oth­er actions like flu­id mech­an­ics to best char­ac­ter­ise the hae­mo­dy­nam­ic beha­viour. The chal­lenge is to cus­tom­ise this heart mod­el to each indi­vidu­al so that we can, for example, test dif­fer­ent ways of treat­ing a patient when he or she has an ill­ness. Based on this con­fig­ur­able mod­el, it is pos­sible to cre­ate a pre­cise repro­duc­tion of the patient’s heart and its beha­viour, both in terms of geo­metry but also in terms of the elec­tro­physiolo­gic­al ellipse or phys­ics. Although a mod­el is only com­prised of the data you put into it, it can nev­er­the­less be used to make predictions. 

We are also cap­able of a multi-scale approach. For the heart, it is pos­sible to change the tool to repro­duce car­di­otox­ic beha­viour. Some drugs will be able to enter heart cells and con­trol the opening/closing of ion chan­nels. Here we enter the field of chem­istry and molecu­lar bio­logy. We can thus gen­er­ate arrhythmi­as of the heart that go as far as tor­sades de pointe [a spe­cif­ic type of abnor­mal heart rhythm], which our soft­ware is able to repro­duce accord­ing to the drug dosage. In this instance we are work­ing on a molecu­lar scale, not just in the field of physics. 

We there­fore need to run sev­er­al digit­al twins of the patient to under­stand their con­di­tion, identi­fy their patho­logy and be able to test dif­fer­ent inter­ven­tion scen­ari­os. While longev­ity has been increas­ing in recent years – accord­ing to WHO fig­ures – healthy life expect­ancy has stag­nated. To over­come this plat­eau, this mul­ti­pli­city will be neces­sary. For example, an inter­ven­tion scen­ario may use a drug or a med­ic­al device, or even both. Anoth­er digit­al twin to test the molecule or med­ic­al device will then be required. This care will be admin­istered in a par­tic­u­lar envir­on­ment: if it is an oper­a­tion, it will take place in a hos­pit­al, but it may be a remote con­sulta­tion or a face-to-face con­sulta­tion in a doc­tor’s sur­gery. We will then need a digit­al twin of the site.

Finally, mak­ing a digit­al twin of the care sys­tem will also require test­ing to under­stand cash­flows and seek the best ways to man­age it. It can even be in 2D – a digit­al twin does not neces­sar­ily need to be in 3D! Even if it is the most inform­at­ive mod­el to illus­trate the tech­no­logy and, in fact, we don’t need to wait until we have a com­plete digit­al twin of the human body to see sol­id results. 

These advances will allow us to move into per­son­al­ised medi­cine and find the right treat­ment for the right patient at the right time. There will also be sev­er­al sec­ond­ary bene­fits. Firstly, the patient will have a bet­ter under­stand­ing of what is going to hap­pen dur­ing the oper­a­tion. This will enable them to become act­ive in their own health­care, thus pro­mot­ing more effect­ive pre­ven­tion. In addi­tion, this pro­gress would free up doc­tors’ time. If a digit­al twin is used to test dif­fer­ent con­fig­ur­a­tions that used to be done by hand, this will free up med­ic­al staff time without under­min­ing the rela­tion­ship with the patient. It also prom­ises bet­ter col­lab­or­a­tion between doctors. 

There is already a high demand from doc­tors for these digit­al twins.

There is already a high demand from doc­tors for these digit­al twins, espe­cially as they com­plain that they spend too much time on admin­is­tra­tion and not enough time with patients. But this is only the begin­ning, and while some are con­vinced of the mer­its of this approach, they are still pion­eers in this field. The need is all the more press­ing now that med­ic­al know­ledge doubles every 70 days. And these doc­tors see the advant­age of being able to pool their know­ledge to train the young­er gen­er­a­tions or to dis­cuss a com­plex case with sev­er­al practitioners. 

One example is the Boston Chil­dren’s Hos­pit­al, where the head of pae­di­at­ric sur­gery, Dr Hogan, uses a mod­el of the heart before oper­at­ing on infants with con­gen­it­al mal­form­a­tions. This digit­al twin of the heart helps him to bet­ter under­stand the child’s ill­ness and allows him to visu­ally explain their child’s con­di­tion to par­ents to build confidence. 

This is proof that the tech­no­logy can be used in advanced med­ic­al prac­tice, but we are still far from wide­spread adop­tion. While the med­ic­al world is keen on innov­a­tions, it also wants them to be proven. How­ever, these tech­no­lo­gies are quite reas­on­able in terms of cost and time. In two- or three-years’ time, they will have greatly evolved in terms of speed, ren­der­ing, res­ults and con­sump­tion of tech­no­lo­gic­al resources, which will encour­age their use through­out the med­ic­al profession.

Interview by Jean Zeid