ausdoc . com . au 9 AUGUST 2024

But the tool needs more testing before widespread use , says Professor Mark Harris .

Rachel Fieldhouse AUSTRALIAN doctors have created a lipid-based risk score to help manage the ‘ troubling group ’ of patients at intermediate risk of CVD .

Based on a routine blood test , the new score provides better cardiovascular risk stratification for those in the intermediate zone than the outdated Framingham-based equation , say the Baker Heart and Diabetes Institute-led team .

They found that for every 100 intermediate-risk patients reclassified using their lipidomic-enhanced risk score ( LRS ), 36 would move up to the high-risk group while 16 would shift down to the low-risk category .

“ This could facilitate the triage of individuals for non-invasive imaging such as coronary artery calcium scoring , fostering precision medicine in CVD prevention and management ,” they wrote in the Journal of the American College of Cardiology .

Emeritus Professor Mark Harris , who was not involved in the research , said that such a tool could vastly improve decision-making for this difficult-to-treat cohort .

“ With low risk , basically the

‘ This test , assuming it ’ s confirmed , would allow us to be a bit more confident .’

focus is on lifestyle interventions … [ and ] with high risk , you ’ re going to be focused on medications and perhaps doing further cardiac assessments ,” said Professor Harris , from UNSW Sydney .

“ This test , assuming it ’ s confirmed with other studies , would allow us to be a bit more confident to know , ‘ Should I be starting either blood pressure- or lipid-lowering pharmacotherapy with this [ intermediate-risk ] patient ?”

The LRS was developed using data from nearly 8000 adults in the Australian Diabetes , Obesity and Lifestyle Study , who were initially free of CVD at baseline and underwent targeted lipidomic analysis .

Its ability to predict CVD events and re-stratify risk was tested against the Framingham equation for external validation in 4000 individuals in the WA-based Busselton Health Study .

“ The reclassified individuals moving up to the high-risk group represented 36 % of all CVD cases initially labelled as intermediaterisk , whereas those moving down to the low-risk group comprised 16 % of

Professor Mark Harris .

all CVD cases initially labelled as intermediate-risk ,” the researchers wrote .

A separate analysis based on a third cohort — the Sydney-based BioHEART study with 1000 adults — showed that the LRS could accurately predict significant coronary artery calcification .

“ These results underscore not only the applicability of the LRS in predicting atherosclerotic markers , but also its potential in guiding decisions for imaging triage ,” the team wrote .

However , they stressed the need for further testing in more diverse populations , given the study cohorts were predominantly of European descent . J Am Coll Cardiol 2024 ; 22 Jul .

Antony Scholefield THE Medicare items for writing and reviewing GP care plans will receive equal rebates from November , but it means that the item for team care arrangements will be scrapped , the Federal Government has confirmed . Reforms to care plans , which are worth over a billion dollars a year to general practice , were announced in last year ’ s federal budget .

Item 723 for team care arrangements ($ 130.25 ) was claimed 2.8 million times over the last financial year .

That item will be scrapped , but item 732 for a care plan review ( currently $ 82.10 ) will rise to the same rebate as care plan item 721 ( currently $ 164.35 ).

The big unanswered question is whether an increase to the review item will mean cutting the care plan item , and if so , by how much .

The changes will mean access to the five Medicare-subsidised allied health treatment sessions per year would be triggered through care plan item 721 .

However , the government says patients would only be eligible for subsidised allied health sessions if their care plan was written or reviewed in the last 18 months .

FROM PAGE 1 keep treating ? No . This is where AI of the future can help .

An algorithm can be used to collect audio from all the consults the patient had with her doctors — the stuff collected by Best Practice ’ s new Lyrebird bot .

This information is then analysed , with highlights played back to the patient ’ s daughter and treating doctor .

This includes all the excerpts of the mother talking about what mattered most to her .

“ You and the daughter listen to these recordings together ,” continues Dr Brender , an internal medicine resident at the University of California .

“ Then you share that another algorithm , trained on seven million patient records , predicts that the mother ’ s chance of walking again is less than 5 %.

“ It also predicts that three [ out ] of four people like her would want to pursue comfort-focused treatment in this situation .” The result ? By analysing the speech content , its context and the prosody , AI could expand advance care planning , Dr Brender argues .

By highlighting explicit conversations about goals and values

AAP

HAL from 2001 : A Space Odyssey .

— along with spontaneous comments about family , hobbies or weekend plans — you can demonstrate how a person has lived .

It is not discussed whether the prosody and content of the typical doctor – patient consult can fully capture the actual complex life of your average patient .

After all , not every patient reveals everything about themselves — even if the medical consult is meant to be the closest the secular world gets to the confessional box .

But Dr Brender makes important points .

He says these recordings would “ humanise patients for the medical team at a time when they no longer resemble the person they had been ”.

“ When patients cannot speak for themselves , we ask surrogates to draw on their knowledge of the patient to help the clinical team make individualised treatment decisions .”

But algorithms collecting “ thousands , millions or even billions of direct observations of a person ’ s behaviour ” might paint a more authentic portrait of the way a person lived , compared with a surrogate whose impression is coloured by psychological , emotional and existential stress .

A good death — could that be the

AI ’ s suggested treatment goal ? The article does not go there . Instead , it warns that machine learning is fallible and can go wrong in the vein of 2001 : A Space Odyssey , which will be bad in cases where clinicians place undue trust in their proffered wisdom .

The algorithm could also make the right recommendation but not be able to explain how or why , Dr Brender adds , which is probably the one area where AI sounds genuinely godlike .

The journal editors issued a note on Dr Brender ’ s article admitting they were “ startled ” by the concept

These recordings would ‘ humanise patients ’.

of AI speaking to decision-makers on end-of-life care goals for those who could no longer communicate .

“ Algorithms using millions of direct observations of a person ’ s behaviour to provide an authentic portrait of the way a person lived ? Yikes !” wrote Dr Deborah Grady , JAMA Internal Medicine ‘ s deputy editor .

But she said the viewpoint was published because it was “ very interesting , somewhat scary and probably inevitable ”. JAMA Intern Med 2024 ; 22 Jul .