Hacking to Helping Our drive to innovate and deliver new services via the QChild system Gavin Bott
Healthy Hearing • Whole of birth popula=on on QLD - Offered • Private and Public • +99% screen rate for whole State since 2006 • Downstream service pathway for referrals
Audiology Family Support Child Hearing Clinics Screening Hearing Implant Deadly Ears Early Interven;on
Customised MS Dynamics CRM 365 • All of business for Healthy Hearing • With all data known from the beginning (2004) • 7 Clinical areas, across 4 programs • Approx. 300 users Statewide • +700,000 Healthy Hearing child records (+ 20,000 Deadly Ears and Hearing • Implant migrated records) 270 record types available (tables), with thousands of individual data items • (fields) 4 th genera=on system. •
• The Deadly Ears Program leads Queensland Health ’ s response to reducing the rates and impacts of middle ear disease and conduc=ve hearing loss for Aboriginal and Torres Strait Islander children across Queensland. • Ear and hearing health is a major public health concern as hearing loss affects how a person interacts, understands and communicates with the world around them.
Aboriginal and Torres Strait Islander children have one of the highest rates of middle ear disease and hearing loss in the world. These children experience it: Earlier : They can contract the disease and hearing loss in their first weeks of life. • Frequently : They suffer from the disease and hearing loss o\en and repeatedly. • Severely : They can develop worse forms of the disease and greater levels of hearing • loss. Persistently : They can experience the disease and hearing loss for longer periods of • =me.
Warraber Island • Poruma Island • Kuin Village • St Pauls Community • Badu Island • Mabuiag Island • Yam Island • Yorke Island • Darnley Island • Murray Island • Saibai Island • Boigu Island •
Can QChild Help Us? “Hey, can you get QChild to display these otoscopy images so ENT’s can look at them outside of being on the trip in person?” • No. Sounds predy hard. Tools don’t like it. • But maybe, it does sound good, I’ll ask. • Yeah, It’s hard, but good, not right now…….
• 3 days of running about • Long hours of Dev work • Short presenta=ons are hard • Take risks, get results
• Review Hack & show what was achieved • Review/re-scope business need – a lot of “new” added • Get signatures and endorsement • Schedule build • Deploy and Go Live Aug 2017
Tele-Otology: Has Value? ENT Work conducted on a this kind of review trip currently lasts a week. Using Tele-Otology will now take 2 days • 5 trips planned in 2017: $60k in resources • 25 Days of ENT ;me • 2500 Tele-Otology reviews • 375 children from a wait list
Tele-Otology: Has Value? The impacts of middle ear disease and hearing loss are substan=al at any age. But in young children it can • affect childhood development – including speech, language and cogni=ve development. In some cases, the recurrent nature of the disease can also lead to permanent hearing loss. Middle ear disease and hearing loss can lead to long-term developmental and learning problems, o\en • seen in the educa=onal and home environments. As children go through early childhood educa=on and into schooling, the disease impacts upon their school readiness, communica=on skills, learning abili=es and educa=onal outcomes. In the home environment, it can affect a child ’ s family rela=onships, social skills and contribute to percep=ons of their behavioural problems. • The impacts of middle ear disease and associated hearing loss can have far-reaching social and economic consequences because they influence the trajectory of children ’ s lives towards adolescence and adulthood, including future employment outcomes.
What again? With machine learning…. • 3 days of running about (bit slower) • Long hours of Dev work (more) • Short presenta=ons (are s=ll hard) • Take risks, get results………. Working proto-type – I hope
2017 Hack: (Q)ICA Goal 1 • Data based screening tool via machine learning Goal 2 • Image based screening tool via machine learning Goal 3 • Unleash QICA on the full dataset
(Q)ICA: Goal 1 Data based screening tool via machine learning • Full QChild integra=on – standard business • 10k exis=ng records to work from • Predic=on of “Needs ENT Review Y/N” • Working proto type from Hack • Can this cut 200 cases in need of review to 100 – do I get to claim another day?
(Q)ICA: Goal 2 Image based screening tool via machine learning • Full QChild integra=on – standard business • 300 images with results to work with as of today • Predic=on: ENT Care Y/N + Urgency • Can it cut finer? Cri=cal condi=ons? • Result: Device agnos=c screening tool • We need new knowledge, data, and =me
(Q)ICA: Goal 3 Unleash QICA on the full dataset • At this point, if goal 1 & 2 are achieved and Live – we kind of have to • +700k records available from 2004 • +60k each year • Associated available data sets of change in QLD readily available – economic, social, etc. • Predic=on of…….what can we find?
Thankyou Healthy Hearing
Recommend
More recommend
