Would be interested in whether people have some examples of algorithms in health - not necessarily AI but this could be included. Stats (on behalf of Minister Curran) is running a preliminary workshop for Central Gov on Friday (Simon and I are attending). Would love to hear what else we could add - have a few already (apologies for the acronyms):
• PBFF
Bound to be some more clinically focused ones? Key risks I think we need to start talking about are:
• Source data to validate algorithm
Seems to make the basis for developing an evaluation framework for the sector? There may already be one - feel free to enlighten us (couldn’t see anything useful in the quick lit search I just did).
Any clinical decision support systems would fit the bill I think.
Also PHOs/DHBs running risk of re-admissions algoritms over particular cohorts.
I would be interested in being involved (going forward) - here are my COPD algorithms:https://www.nature.com/articles/npjpcrm201411 https://www.nature.com/articles/srep44702 https://www.nature.com/articles/srep31893
I’m creating an asthma risk score algorithm - fit for a learning health systems approach - the algorithm protocol has just been published:
Introduction Asthma has a considerable, but potentially, avoidable burden on many populations globally. Scotland has some of the poorest health outcomes from asthma. Although ambient pollution, weather changes and sociodemographic factors have been...
There are 1000’s of algorithms but few are adopted in clinical practice.
This Viewpoint describes 4 major barriers to useful risk prediction that may not be easily overcome by new methods in machine learning and, in some instances, may be more difficult to overcome in the era of big data.
Good article on implementation and adoption challenges (CVD):
PREDICT (Uni of Auckland CVD risk tool - Rod Jackson) is world leading in this regard:https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(18)30664-0/fulltext https://jamanetwork.com/journals/jama/fullarticle/2683125?utm_source=silverchair&utm_medium=email&utm_campaign=article_alert-jama&utm_content=olf&utm_term=052918 : title=“Big Data and Predictive Analytics: Recalibrating Expectations | Clinical Decision Support | JAMA | JAMA Network” : description=“This Viewpoint describes 4 major barriers to useful risk prediction that may not be easily overcome by new methods in machine learning and, in some instances, m”], opengraphobject:[351561245748891 : https://academic.oup.com/qjmed/article/103/10/727/1552364 : title=“Barriers to cardiovascular disease risk scoring and primary prevention in Europe | QJM: An International Journal of Medicine | Oxford Academic” : description=“Abstract. The prevalence and burden of cardiovascular disease (CVD) is high, and it remains the leading cause of death worldwide. Unfortunately, many individua”], opengraphobject:[351561245748958 : https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(18)30664-0/fulltext : title=“” : description=“”], opengraphobject:[351561245749009 : https://www.nature.com/articles/npjpcrm201411 : title=“Development and validation of a model to predict the 10-year risk of general practitioner-recorded COPD | npj Primary Care Respiratory Medicine” : description=“Article”], opengraphobject:[351561245749031 : https://www.nature.com/articles/srep31893 : title=“Projecting the COPD population and costs in England and Scotland: 2011 to 2030 | Scientific Reports” : description=“Article”], opengraphobject:[351561245749046 : https://www.nature.com/articles/srep44702 : title=“External validation of a COPD prediction model using population-based primary care data: a nested case-control study | Scientific Reports” : description=“Article”], opengraphobject:[351561245749243 : http://bmjopen.bmj.com/content/8/5/e023289 : title=“” : description=“”]
Volume forecasting models, Capplan bed demand projections, Midland trauma database
Great - think we have focused on a couple of more detailed examples - CPAC and another - @simon.ross completed the requirements for MBIE.
For Stats NZ (on behalf of digital Ministers) - yes, we focussed on examples where algorithms had a direct influence on health consumers (CPAC - elective surgery and interRAI - needs assessments) in the first instance. This was per the guidance of what was required for the first briefing.
Got MBIE on my brain at the moment
