New ma­chine learn­ing tool to pre­dict hu­man BP: Study

Oman Daily Observer - - SCITECH -

RE­SEARCHERS, in­clud­ing one of an In­dian-ori­gin, have devel­oped a wear­able off-the-shelf and ma­chine learn­ing tech­nol­ogy that can pre­dict an in­di­vid­ual’s blood pres­sure and pro­vide per­son­alised rec­om­men­da­tions to lower it.

When doc­tors tell their pa­tients to make a lot of sig­nif­i­cant life­style changes ex­er­cise more, sleep bet­ter, lower their salt in­take etc. it can be over­whelm­ing, and com­pli­ance is not very high, Sujit Dey, Pro­fes­sor, Depart­ment of Elec­tri­cal and Com­puter En­gi­neer­ing at the Univer­sity of Cal­i­for­nia in the US, said in a state­ment.

“What if we could pin­point the one health be­hav­iour that most im­pacts an in­di­vid­ual’s blood pres­sure, and have them fo­cus on that one goal in­stead,” Dey said.

The study af­firmed the im­por­tance of per­son­alised data over gen­er­alised in­for­ma­tion as the for­mer was more ef­fec­tive.

The team col­lected sleep, ex­er­cise and blood pres­sure data from eight pa­tients over 90 days. Us­ing ma­chine learn­ing and the data from ex­ist­ing wear­able de­vices, they devel­oped an al­go­rithm to pre­dict the users’ blood pres­sure and show which par­tic­u­lar health be­hav­iours af­fected it most.

Newspapers in English

Newspapers from Oman

© PressReader. All rights reserved.