Powered by glolg
Display Preferences Most Recent Entries Chatterbox Blog Links Site Statistics Category Tags About Me, Myself and Gilbert XML RSS Feed
Wednesday, Feb 27, 2019 - 22:35 SGT
Posted By: Gilbert

Medical Notes

With so much recent news on health and medicine (without even considering the latest data leak on HIV), I figured it a swell time for a theme post on the topic. Now, from personal experience, the local medical scene has been relatively well-managed, if with a heavy dose of self-reliance: see for example the requisitioning of relatives' Medishield funds before drawing upon the public coffers. Tangentially, official stats have finally revealed that local elderly are receiving just S$355 a month from their CPF, which likely implies that Roy Ngerng's (remember him?) assertion that "CPF payouts are inadequate, and among the worst in comparable developed nations" is probably true, as calculated here back in 2014. But it's not as if objective truth was ever much valued for its own sake.


Defensive Medicine

A recent case against a doctor for not informing his patient of possible complications has brought on first a S$100000 fine, and then a backlash against its encouragement of defensive medicine - the mandating of procedures that result in a poor trade-off in terms of patient and physician benefit. It so happens that, if you screen for everything, your chances of finding something is pretty good, and that something is more often than not, a false alarm. The lesson then, it seems, is to be fully cognizant of what these trade-offs are; is ordering a S$1000 test for a hundred thousand citizens to potentially save one of them from paralysis responsible? It's easy to say that no expense should be spared with health, but when resources are limited in reality, some form of triage is unavoidable. Further, tests aren't perfect, and come with their own complications...


Gender Quotas

A number of Japanese medical schools have recently been revealed to have rigged their entrance exams to favour men (and those related to alumni). As a punishment, their education ministry has withheld subsidies for Tokyo Medical University, and one expects, the rest of the guilty.

We might at this recall the local quota of having only one-third of the medical intake be female, which was in effect as recently as 2002. Just to play male chauvinist pig's advocate, such quotas might actually be somewhat logical, from a wider societal perspective. Consider a country that has been estimated to require an influx of 200 new doctors annually. At the same time, far in excess of 200 students of both genders are eager to enter training. However, from decades of experience, while about 90% of male doctors will continue in the profession to retirement age, only 65% of female doctors will do so for various reasons, but most prominently due to entering the family way. Assume that all students are of identical medical potential. What selection strategy (in terms of gender ratio) would then result in minimal wasted training resources?

Framed as such, the solution is: all recruited students should be male; ceteris paribus, a male student can be expected to be 90% efficient in terms of service rendered, against 65% for a female student. Put another way, if the country needs 200 doctors annually, it needs to train about 220 male doctors, 310 female doctors, or about 265 total doctors assuming a 50:50 gender ratio (yes, I'm aware that when the retirement takes place matters, this is for illustrating the concept). In practice, not all students are created equal, and so it would be worthwhile to take some of the best female recruits; I expect that the past LKY-led administration settled upon the one-third quota after doing their sums.

Such reasoning likely won't fly as far today, of course - see the Japanese example - and in any case, local shortages have proven to be very pluggable with foreign imports.


Drug Prices

WaPo reluctantly agrees - "apparently accurate".

Shifting our attention over to the United States, the White House's top A.I. adviser - GEOTUS Dr. TRUMP himself - has been successfully prosecuting a War on Drug (Prices) and Big Pharma, after some teething troubles (and yes, Big Pharma probably deserves it; consider Goldman Sachs: "is curing patients a sustainable business model?"). But seriously, perhaps less subtlety and more table-hammering is the way to go in these negotiations; have you heard of the White House's proposal to link drug prices to an international index - as with other nations' price caps - in the mainstream FAKE NEWS? Thought not.

[N.B. And on Goldman Sachs, they're currently being sued by the Malaysian government over 1MDB, after UMNO's historic upset last year. Those wanting the inside track might refer to Tom Wright's Billion Dollar Whale, which covers, among other salacious details, how part of the cash ironically went towards filming The Wolf of Wall Street.]


Show Me The Evidence

The fact that medicine has to be specified as "evidence-based" may strike some as strange - like, isn't medicine a science? You don't hear researchers debating "evidence-based physics", do you? Well, even modern medicine appears as much an art as a science, with senior specialists' opinions oft respected due to their unique experiences - which aren't easily quantifiable and codifiable into textbooks, it seems. The status of hard data and evidence has tended to be kinda iffy in medicine - for every John Snow celebrated for tracing cholera outbreaks by intelligent data mining and analysis, there's a Semmelweis driven to ruin by snooty hide-bound colleagues.

On this, we might examine the case of weed (a.k.a cannabis, marijuana). The momentum towards its legalization and normalization is only building up pace, with the World Health Organization recommending its rescheduling to Schedule I, equivalent to the widely-accepted morphine. The UK has conceded its medicinal effects, while Thailand and South Korea have promptly allowed its import for medical purposes. More generally, there has been much research on heroin and opioid substitution with cannabis, what with the ongoing opioid epidemic in the States (which, given their source, can't help but feel like long-deferred comeuppance on China's part; thankfully, GEOTUS has convinced them to control fentanyl exports while boosting weed, but somehow the mainstream FAKE NEWS sounds like it doesn't want him to succeed, as usual)

Now, evidence can be nuanced, and according to our Law and Home Affairs Minister, the acceptance of unprocessed cannabis for medical use is bunkum, and driven by financial gain. The CNB has dutifully backed him up with "think of the children" Youtube ads - which haven't been too persuasive to local Redditors, who were only too aware of similar evidence with alcohol and some TCM being ignored. Definitely, I can't see them allowing recreational weed anytime soon, and there are historical scars to consider, but this has the feel of a tide that isn't turning back. The most interesting implications, I'd gather, would be on tourism - if weed consumption becomes widespread both in North America and SEA, are the authorities really gonna arrest, say, the 10% or 20% of tourists that are stoners? Methinks they might become more flexible on evidence then...


A.I. vs. M.D.

Artificial intelligence seems the in-thing for now, and the advances just keep coming, what with OpenAI withholding their GPT2 text generation system because it's too good (personally, it's a bit of an exaggeration - yes, they can sorta mimic generic news columns, but that's not too overwhelming given falling journalistic standards). The rise of A.I. has also extended to medicine (a personal professional interest), with a number of well-publicized papers and demos in the past several years plugging human-level diagnostic performance in various fields: ophthalmology, radiology, dermatology, etc. Earlier this month, a US-China collaboration announced a system for diagnosing common childhood conditions from multiple input modalities. Slightly further out of left field, there's mind reading and physiognomy; one also suspects A.I. having a role to play in gene editing, which seems quite popular in truth.

Faced with this mounting evidence of A.I.'s capabilities, it is only natural that discussions about A.I. possibly replacing - if not all, then a good chunk of - doctors, have come to the fore. Those coming from the tech and business side don't see the replacement taking too long, it seems - computer scientist Kai Fu Lee predicts thirty years, while venture capitalist Vinod Khosla has radiologists becoming obsolete in five years, a bold sentiment more or less echoed by deep learning pioneer Geoffrey Hinton.

The radiologists (and other physicians) aren't going down without a fight, definitely. From my reading of their opinions on various forums and subreddits, very few doctors believe their profession under appreciable threat. Other than technical concerns with adversarial attacks, hidden training biases and natural language understanding (which I believe still lags some way behind imaging applications, despite OpenAI's hype), a few popular arguments against A.I.'s proliferation in medicine are often repeated: legal implications (who's responsible?), the presence of other data not typically included in A.I. modelling, the importance of a doctor's soft skills (bedside manner), and ultimately, acceptance by physicians themselves.

On A.I. models not currently being able to consider the full gamut of inputs (e.g. physical examination notes, labs, other patient background) and explain the underlying scientific basis for decisions, I'd say that there really is little preventing A.I. from improving in these respects; as such, the assertion that mid-level practitioners will be more affected is, to me, kinda doubtful. Indeed, it's entirely possible that - assuming legal issues are sorted out - the analysis is farmed out to an A.I., with the "human touch" supplied by cheaper consultants; there really isn't much correlation between a doctor's medical competency and his likeability, after all (see: Gregory House, M.D.)

No, I'd say that physicians' greatest advantage here would be the medical profession's regulatory powers - they can, in many cases, simply veto the use of A.I. tech. Medicine is, after all, one of the very few guild systems that remain. Most countries have a ruling association that restricts memberships through controlling licensing and available training slots, which has largely preserved the prestige - and wage levels - of the profession to the present day. However, this is not an impregnable defence. Consider the case of optometry; consumers have learnt to purchase glasses and contacts online after getting a prescription, eating heavily into optometrists' expected income. While there have been attempts by optometrists to stymie online transactions out of patient care concerns, these are unlikely to work if the cost savings are large enough - which they often are. Note that fervent campaigning against online retailers didn't work out for Sim Lim Square either...


Meanwhile in China
(Source: mobihealthnews.com)


Shanghai's A.I. powered unstaffed medical clinic booths might then be a peek into the future. Let's face it, most consultations are boringly routine; blood pressure, tongue depressor, stethoscope to chest. This costs twenty to thirty bucks here, possibly a lot more in the USA without insurance. Is there a market for having patients wrap the blood pressure cuff over their biceps by themselves, and hold a stethoscope to their own chest - perhaps guided by an automated laser pointer - while being questioned by an A.I. diagnosis chatbot? If it costs just a couple of dollars, I'd say - heck yes. And, you might not even need the A.I. for this; I'd gather there would be more than a few living and breathing general practicioners in less-developed countries willing to do some telemedicine.

It's the same story that has brought down many other craft professions - divide the task up, and mercilessly compete away supernormal profits on the parts that can be automated, by A.I. or otherwise (i.e. assembly-line manufacturing). But given that it'll probably make healthcare much more affordable, I'm not sure if such a shakeup of the medical profession would even be unpopular with the population at large. Open Health, as with Open Software? Let's see if Google is feeling generous.



comments (0) - email - share - print - direct link
trackbacks (0) - trackback url


Next: Intention Manifestation


Related Posts:
On Economics
"Science" & Facts
Corona Update (Part II)
Of Logic, Looting, Twitter & Disinformation (Part II)
Of Logic, Looting, Twitter & Disinformation (Part I)
Back to top

Copyright © 2006-2025 GLYS. All Rights Reserved.