Series 12: Improving healthcare through machine learning and artificial intelligence

Improve­ment and Inno­va­tion Show­case 35: Diag­nos­tic Tools for Gen­er­al Med­i­cine Patients with Short­ness of Breath

Iain Bertram, Data Sci­en­tist, Com­mis­sion on Excel­lence and Inno­va­tion in Health

Machine learn­ing is a tech­nol­o­gy that’s rapid­ly becom­ing piv­otal to health­care to improve the accu­ra­cy of diag­noses, per­son­alise health care, and find solu­tions to prob­lems. ML can assist clin­i­cians in iden­ti­fy­ing anom­alies, pat­terns, and trends while also help­ing to reduce human error.

In this episode, Iain shares his work with the CEIH to devel­op a machine learn­ing mod­el for diag­nos­ing gen­er­al med­i­cine patients with short­ness of breath.

Iain is a Pro­fes­sor of Par­ti­cle Physics and worked at Lan­cast­er Uni­ver­si­ty from 1999 ‑2022. Before join­ing The CEIH Iain spent his time using machine learn­ing on extreme­ly big data sets to inves­ti­gate the fun­da­men­tal prop­er­ties of the Uni­verse, such as the struc­ture of the pro­ton, ​“Why is the uni­verse made of mat­ter instead of anti­matter?”, and is there some­thing weird lurk­ing just beyond reach. Iain is now using this expe­ri­ence to use machine learn­ing to improve health out­comes in South Australia.

Improve­ment and Inno­va­tion Show­case 36: AI’s Influ­ence on Health­care, a Dis­rup­tive Force

Lucie Marsh-Smith, Senior Man­ag­er — Dig­i­tal Design, Inno­va­tion and Change, South­ern Ade­laide Local Health Network.

The health­care land­scape is in the midst of a dis­rup­tive trans­for­ma­tion, all thanks to the infu­sion of Arti­fi­cial Intel­li­gence (AI). Join us as Lucie delves into the ever-evolv­ing syn­er­gy between AI and health­care. Togeth­er, we’ll explore the enor­mous poten­tial, cur­rent appli­ca­tions, and future prospects AI brings to this vital industry.

Lucie has a com­pre­hen­sive and diver­si­fied back­ground in Dig­i­tal Health with over sev­en years of hands-on expe­ri­ence in Dig­i­tal Health SA and CALHN.

Lucie’s career tra­jec­to­ry includes a note­wor­thy stint of 11 years as the Direc­tor of Tech­nol­o­gy at Blue Box IT in the Unit­ed King­dom, where she honed her exper­tise in lead­ing tech­nol­o­gy-dri­ven initiatives.

Lucie is cur­rent­ly pur­su­ing a Ph.D. pro­gram at the Uni­ver­si­ty of South Aus­tralia, with a research focus on Dig­i­tal Inno­va­tion and AI’s piv­otal role in rev­o­lu­tion­is­ing health­care on a glob­al scale.

Improve­ment and Inno­va­tion Show­case 37: Nat­ur­al Lan­guage Pro­cess­ing to Pre­dict Hos­pi­tal Dis­charge, Deriva­tion, Val­i­da­tion and a Case for Implementation

Samuel Gluck, Med­ical Admin­is­tra­tion Reg­is­trar, North­ern Ade­laide Local Health Network

The abil­i­ty to pre­dict when a patient is like­ly to be dis­charged will help clin­i­cal teams to ensure the patient is ready. Sam’s solu­tion uses RAH data to derive an NLP algo­rithm, val­i­dat­ed in both the RAH and QE, that pre­dicts the like­li­hood of dis­charge with­in 48 hours. Sam’s team is con­duct­ing an imple­men­ta­tion study at RAH and QEH to assist with iden­ti­fy­ing patients who will be ready for dis­charge at the weekend.

Sam Gluck is a duel RAC­MA and CICM trainee. He grew up in Wales and trained in Cam­bridge, under­tak­ing anaes­thet­ic train­ing in the NHS pri­or to emi­grat­ing to Aus­tralia in 2013. He has just com­plet­ed a PhD in the use of pas­sive smart­phone data in the mea­sure­ment of patient out­comes. Sam is well pub­lished in machine learn­ing and nat­ur­al lan­guage pro­cess­ing and when not work­ing he can be found restor­ing a small part of the Ade­laide Hills to native bush­land with his wife and 2 young sons.

Improve­ment and Inno­va­tion Show­case 38: Statewide Deep-Learn­ing Mod­els to Esti­mate Length of Stay

Alex Al-Saf­far, Data Sci­en­tist, South­ern Ade­laide Local Health Network.

Hos­pi­tal length of stay (LoS) of patients is a cru­cial fac­tor for the effec­tive plan­ning and man­age­ment of hos­pi­tal resources. There is con­sid­er­able inter­est in pre­dict­ing the length of stay in order to improve patient care and increase ser­vice efficiency.

Alex has devel­oped a tri­par­tite deep learn­ing mod­el for statewide LoS Esti­ma­tion at the time of patient admis­sion. The mod­el per­for­mance exceeds a mod­el that unre­al­is­ti­cal­ly uti­lizes the DRG of the patient. The mod­el is re-trained for every ser­vice of SA Health to pro­vide LoS for all patients with zero input from any clinician.

Dr Alex com­plet­ed his PhD on Data-dri­ven tech­niques for bio­med­ical elec­tro­mag­net­ic imag­ing at The Uni­ver­si­ty of Queens­land in 2021. He has a patent imag­ing neur­al net­work and is cur­rent­ly work­ing with SAL­HN as data-sci­en­tist. Alex’s inter­est span math­e­mat­ics, sta­tis­tics & pro­gram­ming and He’s the author and main­tain­er of mul­ti­ple Python packages.