Digital twins and prevention: a new era in medicine

Digit­al twins are gain­ing ground in the indus­tri­al world. The doors that this kind of tool opens up allow it to be exten­ded into the field of health­care; in par­tic­u­lar with the use of digit­al lungs for per­son­al­ised care. Being able to mod­el an organ, per­son­al­ise this mod­el using patient data, and sim­u­late its func­tion­ing could lead us to plan, anti­cip­ate and pre­dict its devel­op­ment and ageing.

After all, as Claire Biot, Vice-Pres­id­ent of the Health Industry at Dassault Sys­tèmes, rightly points out, “today, by hav­ing the digit­al twin of a Rafale, as well as lots of flight data – the equi­val­ent of bio­mark­ers in aero­naut­ics – we can carry out pre­dict­ive main­ten­ance in oper­a­tion. This means mov­ing from the digit­al twin of a concept, the air­craft, to the per­son­al­ised digit­al twin of each Rafale in operation.”

This raises the ques­tion: what role can digit­al twins play in pre­vent­ive health­care? A ques­tion that was at the heart of the dis­cus­sions at the 5^th sem­in­ar of the Poly­tech­nique health group, coordin­ated by the Centre de recher­che en ges­tion de l’École poly­tech­nique (IP Par­is), and led by Étienne Min­vi­elle (CNRS Research Dir­ect­or)¹.

Health pre­ven­tion is often seen as a sub­ject that was neg­lected until recently. Stan­ley Dur­rle­man, research dir­ect­or at Inria, is quick to point this out. “Talk­ing about pre­ven­tion is still uncom­mon in France, because first and fore­most we have a care sys­tem rather than a health sys­tem.” In his opin­ion, the cur­rent motto in research is more along the lines of “bet­ter to cure than to prevent.”

At Dassault Sys­tèmes, digit­al twins have been in use for some time. “Our ambi­tion is to help people live health­i­er lives at a cost that is sus­tain­able for health­care sys­tems, and we place digit­al twins at the heart of our strategy,” explains Claire Biot, “because they make it pos­sible to rep­res­ent the invis­ible and get sev­er­al dis­cip­lines to work togeth­er.” A new pro­ject has been launched: “Med­iTwin was launched by Dassault Sys­tèmes to cre­ate dif­fer­ent digit­al twins in three thera­peut­ic areas: onco­logy, neur­o­logy and car­di­ology,” she explains. “These digit­al twins will primar­ily be used for treat­ment or dia­gnos­is, guid­ing med­ic­al decisions”.

There are two types of pre­ven­tion: primary pre­ven­tion (put in place to avoid catch­ing the dis­ease) and sec­ond­ary pre­ven­tion, which stops or slows down the devel­op­ment of the dis­ease. Primary pre­ven­tion applies in par­tic­u­lar to com­mu­nic­able dis­eases, with con­doms and quar­ant­ine as examples. It is there­fore cer­tain that, for cer­tain dis­eases, primary pre­ven­tion is com­plic­ated. The second form of pre­ven­tion has a num­ber of con­straints, in par­tic­u­lar the need for rap­id dia­gnos­is and long-term patient man­age­ment. “How can I fol­low patients for dec­ades to see if, in the end, my pre­ven­tion meas­ure was effect­ive?” won­ders Stan­ley Dur­rle­man. This is where a digit­al mod­el­ling sys­tem that faith­fully sim­u­lates the func­tion­ing of a per­son­al­ised bio­lo­gic­al sys­tem using patient data comes into its own.

When it comes to illus­trat­ing the com­ple­ment­ar­ity between pre­ven­tion and the pre­dict­ive algorithmic sys­tems pro­posed by digit­al twins, Stan­ley Dur­rle­man con­siders Alzheimer’s dis­ease to be the prime example. Alzheimer’s dis­ease is cur­rently not far behind can­cer as one of the biggest areas of med­ic­al research. “Des­pite colossal invest­ments in research, we haven’t had a pos­it­ive clin­ic­al tri­al for this dis­ease for 20 years,” explains the research­er. “And prac­ti­tion­ers still have vir­tu­ally noth­ing to pre­scribe to their patients” [editor’s note: the drug in ques­tion is cur­rently only avail­able in the United States, and the applic­a­tion to mar­ket it in Europe is still being examined by the authorities].

This treat­ment tar­gets the accu­mu­la­tion of amyl­oid plaques in the patient’s brain, which begins 10–15 years before the first symp­toms of the dis­ease appear. “How­ever, the effects of the treat­ment on the patient’s cog­nit­ive decline remain rel­at­ively minor,” he admits. “The change in our treat­ment plan is sub­stan­tial. We are no longer look­ing for drugs to cure the dis­ease, but rather to pre­vent it.”

It is there­fore import­ant to under­stand how this dis­ease devel­ops and what the first signs are, in order to make an early dia­gnos­is. To this end, Stan­ley Dur­rle­man’s research team has estab­lished pre­dict­ive mod­els – digit­al twins – that can determ­ine the course of the dis­ease in dif­fer­ent people. After present­ing this mod­el, Claire Biot asked: “Are we ready for mass-test­ing for Alzheimer­’s dis­ease?” Stan­ley Dur­rle­man replies that there is still pro­gress to be made. “For a non-com­mu­nic­able dis­ease, it’s more accur­ate to talk about screen­ing,” explains the research­er. In gen­er­al, for this type of ill­ness, the patient will a doc­tor fol­low­ing the first symptoms.

Iden­ti­fic­a­tion there­fore depends on the patient tak­ing action, and this first step must not be missed. “Giv­en the issues we have already dis­cussed, I think there is still a lot of work to be done to improve the detec­tion of cog­nit­ive dis­orders in the gen­er­al pop­u­la­tion,” he con­cludes. One of the areas to be developed would be the iden­ti­fic­a­tion of bio­mark­ers spe­cif­ic to the devel­op­ment of Alzheimer­’s dis­ease. How­ever, just as the accu­mu­la­tion of amyl­oid plaques in a patient’s brain is not always syn­onym­ous with the first signs of the dis­ease, pre­ven­tion will always depend on our abil­ity to predict.

These find­ings con­firm that, even if ques­tions remain, the digit­al twin is likely to play an import­ant role in the com­ing years. It could her­ald a new form of health pre­ven­tion, more algorithmic, more per­son­al­ised and offer­ing pre­vi­ously unima­gin­able detec­tion capabilities.

Pablo Andres