Using social networking and ar1ficial intelligence to collect FHx - PowerPoint PPT Presentation

Using social networking and ar1ficial intelligence to collect FHx Brandon M Welch, MS, PhD Assistant Professor Medical University of South Carolina

Family history is important

Cancer runs in families

Life6me risk for breast cancer 70% 60% 50% 40% 30% 20% 10% 0% Average risk Smoker Drinker 1 FDR 2 FDR BRCA2 3 FDR BRCA1 Average risk Smoker Drinker 1 FDR 2 FDR BRCA2 3 FDR BRCA1

Most people have cancer in the family % of US populaPon 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Smokes Close family Extended family (parent, sibling) (+ grandparents, aunts/uncles) Smokes Breast cancer Colon cancer Prostate cancer Lung cancer

It’s in the family, but where’s the gene?

VUS = fly in gene6c tes6ng ointment

VUSs are growing

Use family history to make sense of VUSs

Collec6on of FHx

<4%

20-30 minutes 2.5 minutes Time needed Time spent

100% Breast cancer 30-90% accurate Mental health 0%

Clinicians don’t have 1me to collect a bad FHx

PMID: 15538320

“painful to use” • Poor user interface design • Complicated and illogical quesPons • System crashes • High user workload 1 in 10 3 of 4 complete FHx never finished

Reimagine FHx collec6on

Leverage the family social network Interoperable with EHRs Provide care recommendations Support clinical research PMID: 25868012

Chatbots

How lik likely ely would you use ItRuns? PMID: 26958272

Phase 1 Phase 2 Under development • MVP complete • Open beta • Pilot studies • HL7 interoperability

FHx risk assessment

Family cancer risk assessment is a me mess Risk for Risk for Lynch Risk for Risk for RAD51 gene syndrome FAP BRCA gene Risk for Li Fraumeni syndrome Risk PancreaPc Risk for prostate Risk for Risk for colon Risk for skin Risk for cancer cancer ovarian cancer cancer cancer breast cancer Wijnen Pancpro MELApro Couch MMRpro Gail Cancer Guidelines MMRPredict Claus 531 statements PREMM1,2,6 55 guidelines published Penn II 11 different organizaPons Myriad Tyrer-Cuzick BRCAPRO BOADICEA BRST

Unify the risk tools Risk for Risk for Lynch Risk for Risk for RAD51 gene syndrome FAP BRCA gene Risk for Li Fraumeni syndrome Risk PancreaPc Risk for prostate Risk for Risk for colon Risk for skin Risk for cancer cancer ovarian cancer cancer cancer breast cancer Unified Hereditary Cancer Assessment Tool (U-CHAT)

Unified Hereditary Cancer Assessment Tool (U-CHAT) Under development

Opportuni6es

Cancer risk

Personalize care

Genome discovery

Research recruitment

Open collabora6on plaTorm ApplicaPon & funcPonality Data collecPon and storage as a pladorm