“Let’s personalise healthcare pathways, too”

Just as the prac­tice of medi­cine is more than pre­scrib­ing med­ic­a­tion, per­son­al­ised medi­cine more than tar­geted treat­ments. Étienne Min­vi­elle, pro­fess­or at École Poly­tech­nique is work­ing on more per­son­al­ised health­care path­ways. His aim is to con­sider patients as a whole, tak­ing into account their social con­text, habits and even beliefs.

“We’re talk­ing about a per­son­al­ised approach through­out the the entire jour­ney of a patient through the health­care sys­tem. From hos­pit­al to home, includ­ing prac­ti­tion­ers such as inde­pend­ent nurses and phar­macists,” Min­vi­elle explains. 

Offer patients the best guidance

The idea is to draw on tools developed in the ser­vice and e‑commerce sec­tors to tail­or cus­tom­er exper­i­ence. How­ever, instead of offer­ing dis­counts based on shop­ping habits, per­son­al­ised care means tak­ing into account the patient’s social determ­in­ants and psychoso­cial needs.

“Once dia­gnosed with a chron­ic ill­ness, a patient will require ser­vices and non-med­ic­al inter­ven­tions to improve their qual­ity of life. This could involve get­ting a pet-sit­ter dur­ing treat­ment ses­sions, help with paper­work to report their long-term con­di­tion, or organ­ising trans­port,” he adds.

A patient’s beha­viour and per­son­al­ity also affect the way they relate to health­care. They might, for example, have a habit of research­ing health-related inform­a­tion online, tend to engage in risky beha­viour, have a pos­it­ive or neg­at­ive out­look, be intro­ver­ted, extro­ver­ted, etc. “All of these factors influ­ence the patient’s response,” Min­vi­elle says. In order to take them into account, he is devel­op­ing patient-centred man­age­ment models.

These tools require data, and AI can facil­it­ate data col­lec­tion and help build rel­ev­ant pro­files. “But in health, much of the data is highly com­plex,” he says. “We’ll need new pro­fes­sion­als, such as coordin­at­ing nurses, to make sense if it.” Con­nec­ted objects, such as pill dis­pensers, could also be used; blood tests could be pre­scribed to check for meta­bol­ites so as to mon­it­or patient com­pli­ance. “These solu­tions must go hand in hand with eth­ic­al con­sid­er­a­tions. How far can we go without intrud­ing on patients’ lives?”

Pre­serving patient autonomy

When it comes to eth­ics, vigil­ance is required on sev­er­al fronts. “Per­son­al­ised medi­cine must not restrict access to care,” Min­vi­elle stresses. “There are risks per­tain­ing to access to per­son­al­isa­tion, as well as minor­ity rep­res­ent­a­tion in our mod­els.” In a sys­tem that adapts patient care path­ways to their spe­cif­ic pro­file, digit­al mod­els could be used to organ­ise care options into broad cat­egor­ies. This would stream­line care path­ways, restrict­ing them in a way that may omit less pop­u­lar pref­er­ences with­in the patient population.

Lastly, per­son­al­isa­tion must not come at the expense of patient autonomy. Min­vi­elle believes that “patients should be able to opt out of care path­way per­son­al­isa­tion, and choose their own per­son­al­isa­tion path­way, even if it is not optimal.” 

Human and arti­fi­cial intelligence

Giv­en what’s at stake, such a pro­gram can­not solely rely on digit­al tools. Human intel­li­gence is required. Min­vi­elle reminds us that “AI can­not yet detect its own errors, where­as a human will imme­di­ately notice if a patient has been giv­en the wrong advice.”

Will this sys­tem save money? “It remains to be seen. On the one hand, it should reduce waste; on the oth­er, we would be cre­at­ing a more com­plex sys­tem of care.” Min­vi­elle agrees that more research is needed, and truly per­son­al­ised care path­ways are still some time off.

“The main dif­fi­culty is bring­ing everything togeth­er”, he says. The indi­vidu­al­isa­tion mod­el is based on research from a num­ber of fields: pub­lic health (integ­rat­ing socioeco­nom­ic determ­in­ants); psy­cho­logy (ana­lys­ing patient beha­viour and demands); and med­ic­al research (examin­ing the effic­acy of treat­ments on vari­ous patient pro­files). Yet “all of these groups are work­ing in isol­a­tion. If we are to devel­op a cul­ture of per­son­al­isa­tion with­in the health sys­tem, we have to demon­strate its feasibility.”