Apple/ Eli Lilly’s bet: Wearable and mobile consumer devices may well help us detect cognitive impairment and dementia

by

__________

Apple, Eli Lil­ly research whether devices can detect demen­tia signs (Health­care Dive):

“Demen­tia, which affects rough­ly 47 mil­lion peo­ple across the globe, costs $1 tril­lion world­wide, accord­ing to the World Health Orga­ni­za­tion. Ear­ly test­ing for the con­di­tion is spo­radic and, when con­duct­ed, it’s often not sen­si­tive enough to detect ear­ly stages of men­tal decline, cre­at­ing an oppor­tu­ni­ty for tech com­pa­nies like Apple to see whether they can turn a profit.

The “rich, lon­gi­tu­di­nal infor­ma­tion” from wear­able and mobile con­sumer devices can be “mined for phys­i­o­log­i­cal and behav­ioral sig­na­tures of cog­ni­tive impair­ment and pro­vide new avenues for detect­ing [mild cog­ni­tive impair­ment] in a time­ly and cost-effec­tive man­ner,” researchers wrote in the paper. The team includ­ed five authors from Apple, five from Eli Lil­ly and five from Evidation.

Data was col­lect­ed from a slew of plat­forms includ­ing an iPhone, Apple Watch, iPad and a Bed­dit sleep mon­i­tor­ing device, then stored on a HIPAA-com­pli­ant Evi­da­tion plat­form. Researchers had access to the data through a secure net­work called a VPN before mod­el­ing … The ini­tial results are promis­ing, but researchers stressed they’re only a “start­ing point” for fur­ther stud­ies on using the con­sumer gad­gets to pre­dict cog­ni­tive decline, which includes Alzheimer’s disease.”

The Study:

Devel­op­ing Mea­sures of Cog­ni­tive Impair­ment in the Real World from Con­sumer-Grade Mul­ti­modal Sen­sor Streams (Pro­ceed­ings of the 25th ACM SIGKDD Inter­na­tion­al Con­fer­ence on Knowl­edge Dis­cov­ery & Data Mining):

  • Abstract: The ubiq­ui­ty and remark­able tech­no­log­i­cal progress of wear­able con­sumer devices and mobile-com­put­ing plat­forms (smart phone, smart watch, tablet), along with the mul­ti­tude of sen­sor modal­i­ties avail­able, have enabled con­tin­u­ous mon­i­tor­ing of patients and their dai­ly activ­i­ties. Such rich, lon­gi­tu­di­nal infor­ma­tion can be mined for phys­i­o­log­i­cal and behav­ioral sig­na­tures of cog­ni­tive impair­ment and pro­vide new avenues for detect­ing MCI in a time­ly and cost-effec­tive man­ner. In this work, we present a plat­form for remote and unob­tru­sive mon­i­tor­ing of symp­toms relat­ed to cog­ni­tive impair­ment using sev­er­al con­sumer-grade smart devices. We demon­strate how the plat­form has been used to col­lect a total of 16TB of data dur­ing the Lil­ly Explorato­ry Dig­i­tal Assess­ment Study, a 12-week fea­si­bil­i­ty study which mon­i­tored 31 peo­ple with cog­ni­tive impair­ment and 82 with­out cog­ni­tive impair­ment in free liv­ing con­di­tions. We describe how care­ful data uni­fi­ca­tion, time-align­ment, and impu­ta­tion tech­niques can han­dle miss­ing data rates inher­ent in real-world set­tings and ulti­mate­ly show util­i­ty of these dis­parate data in dif­fer­en­ti­at­ing symp­to­matics from healthy con­trols based on fea­tures com­put­ed pure­ly from device data.

The Study in Context: