Hacker News Reader

Comment by amalgamated_inc | Hacker News Reader

Would be cool if we had wearable patches like Continuous Glucose Monitors, but monitoring for all kinds of diseases. If these were ubiquitous, we'd detect diabetes, Alzheimer's, many types of cancers, maybe heart disease years before they showed symptoms.

View on HN

Easier to just screen for this during annual or biannual physicals with a blood draw, no? Would be sufficient for all except the most aggressive and fast moving cancers.

“Wouldn’t it be cool” and “Wouldn’t it be easier“

Well, it would be cool to get more real-time feedback, where possible.

And sure, somethings will be easier when blood is drawn.

So, “yes” to both questions

There are lots of things I'd really love to have a graph instead of a single data point for. The obvious candidate we already have a solution for is glucose, but I'd also love figure out how well my hormone replacement is working, because my current level is care is "oh, that's a bit high, you took that 1-2 hours before the test huh" (no instructions were given) once a year.

> There are lots of things I'd really love to have a graph instead of a single data point for.

I couldn't agree more.

Fortunately Apple Health is really good at this [1].

[1]: https://www.apple.com/ios/health/

> Easier to just screen for this during annual or biannual physicals with a blood draw, no?

Short answer: no.

You’re not going to be screened for certain things on your annual unless your age or condition warrants it.

Unless you’re in a high risk group, they’re not going to recommend a colonoscopy if you’re under 50-55 years old. That could change if enough data has been collected to suggest otherwise.

That recommendation changed ~4.5 years ago to be 45 and older.

https://www.cdc.gov/cancer/colorectal/basic_info/screening/i....

And earlier still if you have a family history or certain bowel diseases. From that page:

However, you may need to be tested earlier than 45, or more often than other people, if you have—

Inflammatory bowel disease such as Crohn’s disease or ulcerative colitis.

A personal or family history of colorectal cancer or colorectal polyps.

A genetic syndrome such as familial adenomatous polyposis (FAP)external icon or hereditary non-polyposis colorectal cancer (Lynch syndrome).

Correct; I was responding to correct a conditional which started "Unless you’re in a high risk group..."

If people don't think something is worth checking for annually, why would they check for it continuously?

If we were at the point that wearable, multi-illness screening patches are affordable and available, would it not follow that annual screens would also be just as available?

> Easier to just screen for this during annual or biannual physicals with a blood draw, no?

8% of the US population (about 26.4 million people) don't have medical insurance [1]; there's no annual physical for them. Unfortunately they only get medical care via the emergency department. I bet a lot of them have smartphones.

BTW, 8% is an all-time low.

[1]: https://aspe.hhs.gov/reports/2022-uninsurance-at-all-time-lo...

> Would be cool if we had wearable patches like Continuous Glucose Monitors, but monitoring for all kinds of diseases.

I think we’re well on the way for this.

Already the Oura ring and the Apple Watch are collecting data to detect arrhythmias and other heart issues. During the height of the COVID pandemic, the Oura ring detected infections prior to people having noticeable symptoms.

Thanks to my Apple Watch, I have years of medical data including things that aren’t normally tracked by most medical professionals in the US such as walking steadiness (a metric for the risk of falling), heart rate variability, resting heart rate and lots more that can give doctors a heads up that something not good is happening.

The problem is, unless you’re part of a very progressive medical practice or in a clinical trial, there’s often no way to get the medical establishment to pay attention to this data or get it into their IT systems.

Older doctors especially are conditioned to basically ignore patient reported (or collected) data that’s not temperature or weight.

The issue with this sort of broad-based testing is specificity. If these tests aren't highly specific then you're going to get all sorts of false positives clogging up the medical system. That can be such a large burden that proper care for sick people gets negatively affected.

In countries where medical care isn't the first step to financial ruin, people have regular checkups all the time. This sort of thing could just be used as a basis for further decision making (as are health apps right now), with only things that need immediate action turning into a "go to the doctor right now".

If a doctor would ignore a positive result if they saw one then they shouldn't order the test!

Which countries are those? There are a few screening and preventative care services which have proven valuable, and are covered in the US at no cost to the patient. But regular check-ups are pointless for most of us.

https://www.healthcare.gov/coverage/preventive-care-benefits...

I worry for the day we have this, but all the measurements needing to be stored on someone else's computer to be able to do the detection.

I use Apple’s Health app; it connects to my medical provider’s Epic system. Epic is like the Oracle or SAP of the enterprise medical records world with 300 million patient records.

All test results are automatically downloaded.

The Health app acts as a repository of data; a user can grant access to 3rd party health apps to analyze specific chunks of data if desired.

Apple makes available onboard machine learning hardware + APIs; using them, apps can at least provide guidance for certain conditions, maybe even before your doctor knows.

I suspect Apple will continue to innovative in this area, especially as they add more sensors to the Apple Watch and AirPods and make better use of the existing sensors.

you'd be up to your ears in false positives, which can often be worse. screening is hard.

having many data points over the past month is arguably better for a physician than having a single data point from a visit

But only if the false positive rates are low enough, and/or the interventions required to follow up a false positive is small enough, and the benefit of earlier intervention in the case of true positives outweigh those risks.

Large scale testing is highly dicey and requires careful consideration because it often takes very low risks from false positives to outweigh the benefits.

We seem pretty great about not having false positives with blood glucose and diabetes. There are of course lots of false positives in other areas. But I'm sure we could dial in many of those type-1/type-2 errors, or just look at the ones where we're confident in the ratio and that there's a net benefit.

Of course there are false positives. Any time you do anything at scale there will be, be it due to precision of equipment, equipment failure, or user error.

When a true positive is rare, the false positives can easily swamp them. E.g. one Canadian paper claims a glucose test for diabetes gives a false positive rate 6 times higher than true positive in the general population [1] (the numbers will of course be much better in a group where risk factors are present).

A challenge is that becoming confident of the ratio is difficult. What do you count? Routine mammography has been scaled back some places because even miniscule false positive rates seemed to rack up enough downsides alongside surprisingly low benefits (you caught things earlier, but the difference turns out to have less impact on survival rates than you might think) when the true positive rate is also tiny.

The end result is that the candidate pool for large scale screening programs where we can be confident of a net benefit is surprisingly small, and it may well be we'll see further declines in use of large scale screening rather than increases, as improved treatment often further cuts into the net benefit of early detection.

[1] https://pubmed.ncbi.nlm.nih.gov/9368868/

There's a whole literature and backstory supporting my statement, if you want to engage do the backing research.

It's not reasonable to demand people do legwork if they want to reply to your naked claims. This is not a forum exclusively for people who are well-versed in the research you're familiar with.

it's not a naked claim; it's a well understood problem in the field that is trivially researchable and this comes up over and over again on HN.

HN is not in "the field," and if it's trivially researchable then you can include links to it yourself. It's not constructive to demand other people are read up on research before interacting with you here.

Reply on news.ycombinator.com