DPV HEALTH
“ Then they ’ re going to turn it on , think it ’ s intelligent , and use it to search through data and pull together answers for their questions – that ’ s an obvious , real benefit for an organisation . But they ’ re not necessarily going to be prepared for the fact that , historically , they ’ ve had really bad practices managing their systems and files .”
“ If you unleash AI and haven ’ t segmented and haven ’ t got good internal controls with role-based access , and haven ’ t locked down those documents , and they ’ re not encrypted , you ’ re going to get all sorts of privacy issues .”
Toal ’ s advice is don ’ t rush into delivering AI without thinking about what the problem is you are trying to solve . He also suggests not using the word AI when working out the problem because it will “ just scare people ”. Instead , he says just talk about the problem and what you want to do better , and then work out how AI can provide a solution for each element .
DPV Health have built a proof of concept predictive AI model to address the fact that around 25 % of clients would fail to attend appointments .
“ That ’ s an instant hit to productivity , and we have massive wait lists because we ’ re publicly funded within a rapidly growing community , to get people in who really need help ,” explains Toal . “ So if the clinician sits there not servicing someone , that ’ s impacting us and the clients ”.
“ The model predicts at the time when clients are making the appointment their likelihood of attending . We ’ ve tested about 91 % accurate on that , so that ’ s been really good , and with a number of other data driven changes those no-shows are now around 14 % – or 8 % if you exclude mental health appointments – down from 25 %.” technologymagazine . com 169