I tried feeding a lot of this into various Als (Kagi gives you access to a few with a nice interface) and I found that they mostly were stupid in ways that made me think.
Perfect summary of why AI is useful for rubber ducking. Doing this with a co-worker you always get some questions that are a few layers too shallow for what you need (because they haven't been thinking about it more than a few minutes), but it always makes you think and really helps.
Even when the duck doesn't answer at all it still often works.
I worked next to another engineer in the early eighties who was working on a different project. At tea break time we got into the habit of sitting opposite each other and explaining the problems we were having. Neither of us put much, if any, effort into trying to solve the other's problems because it often turned out that simply attempting to formulate the problem in a way that would make sense to the other party was enough to reveal the answer or at least a promising idea to try.
This. Previous cow-orker of mine that was doing remote-only like 10+ years ago at some other company got an actual yellow rubber duck sent to him together with his company laptop when he joined.
I also like to do this when debugging. I'll just write out whatever my current hypotheses are about what the bug is and I adjust as I go, as I disprove or make them more precise. I would usually just do it right on the ticket. It both helps because it provides the "rubber duck" and in the end I just submit the comment and I have a root cause analysis right there.
Unfortunately it's quite rare to see this from anyone and asking them to provide a root cause analysis is usually fraught with peril. I don't quite get why. It's been working awesomely.
And like OP says, AI hasn't made it better in the sense that not everyone can use AI (or cow-orkers) in that way. They just believe what the AI tells them and they literally won't reverse engineer that protocol in 24h ;)
> A few times I thought they had "cracked the case" but actually they just made me waste time.
Despite trying various models and services for years now, I've come to the conclusion that the current iteration of AI services hurts more than it helps. I spend more time prompting, reading pages of babble, and trying to find signal in the noise that I would just chilling out and thinking through the problem.
My friend went to see the doctor. A very attractive nurse took his blood pressure, which turned out to be off the charts. She goes out, comes back 5 minutes later and repeats, same results. Goes out and 10 minutes later and old bearded doctor comes in and repeats the tests.
Doctor ends up telling him that he was hoping it was just the attractive nurse, but no. He'd be starting medication today.
I have an intense phobia of dentists; if I were wearing a constant monitor, I bet you could tell when I was at the dentist just by watching the blood pressure and heart rate spikes. (You'd have to find some way of differentiating them from me being in a car accident, or being attacked by a werewolf.)
I wonder if it would show up on the heart rate (or heart rate variability) tracking from an Apple Watch or similar. My Garmin picks up stressful events all the time.
I have a mild fear of needles. Whenever my wife does acupuncture on me, I am quite uneasy in the beginning, looking away but still I end up covered in a bit of sweat till she finishes putting needles on all the points (and sometimes they are properly weird places I wouldn't imagine sticking tiny needle into... doesn't help). Don't monitor the heart rate/pressure but it must jump up significantly
My wife developed an autoimmune condition a few years ago that causes her to need to see a specialist every few months. For some reason, the hospital we've been going to uses these automated blood pressure readers that literally need to get wheeled around because they're four feet tall with a digital display, and during one of the early visits she had because of one of these conditions, the device bugged out or something and it kept squeezing her arm without releasing the pressure, and the person taking the reading didn't know how to disengage it manually and ran out of the room to go and find someone to help them without taking the time to rip off the velcro band on my wife's arm or anything. Ever since then, I literally get stressed every time my wife's blood pressure gets taken from one of those devices even though I'm not even the one who's getting measured!
I had a doc appointment a couple of weeks ago. My wife was driving and we were running late, so I was getting stressed about missing the appointment on top of being stressed about going to see the doctor. There was a lot of traffic so I decided to get out at a red light and run to my appointment while my wife parked the car. I only barely made it and the first thing they did was take my pressure. They took it twice, because at the first they were like, this can't be right. When I explained, they asked if I had a blood pressure monitor at home so I could take my own pressure at a more quiet moment.
The guidelines for taking blood pressure state you should be sitting quietly for 5 minutes beforehand, something that almost never happens at the doctor's.
I've never been to a doctor where they waited the appropriate amount of time. It's always almost immediately as you sit down. This has caused a lot of problems for me. I learned most insurance companies require a doctor hold you until your blood pressure is "normal" (where normal is < 140/90 for their purposes). This resulted in several times where I was kept in an office with nurses coming in and out every 3 minutes to repeatedly take measurements which simply made it worse.
Blood pressure is highly variable even moment by moment. A single data point at your yearly visit is basically meaningless. Even your posture can shift it. Legs in a different position? Too much tension in one arm over the other? Seated awkward? Have gas? Unfortunately, it's also very difficult to get a 24 hour monitor unless you have another suspected condition (such as kidney problems).
I probably have this, I've had several specialists take my blood pressure during consultations and express concern. I was prescribed a 24 hour ABMS (ambulatory blood pressure monitoring system, ie. wearable monitor) test, and came back with normal (if slightly high) readings.
This prompted me to do some research, and it turns out there's very strict guidelines for how to 'accurately' measure blood pressure, almost none of which were followed by most specialists. There's a laundry list of things that can each increase your blood pressure by 10-15mmHg (eg. drinking water, sitting cross legged, recent physical exercise, not sitting with the right posture, not sitting for at least 15 minutes).
Given this lackadaisical approach to measurement, I seriously question the diagnostic validity of blood pressure tests 'in the wild'.
White coast hypertension isn't just about a clinical setting, it's more generally about the worry about the results of having your BP measured. [EDIT] OK, well maybe there's another type of hypertension which is related to anxiety about a high result regardless of the measurement setting.
I have to take daily BP measurements during titration for ADHD medication. (Using an _A&D UA-611 Plus_ machine at home.)
I can put the cuff on my arm and sit at my desk for 20 minutes to be nicely rested and calm, and then take 5 different measurements with a few minutes between each one. They'll vary quite wildly (anything from, say, 115/75 to 135/90) despite not moving between measurements or having any reason to be more or less agitated. I generally just ignore the low/high outliers and average the others. There's no pattern either, sometimes the outliers are first, sometimes last.
Also it's not just the monitor I have at home, the same is true of a probably more trustworthy machine in a clinical setting. I mentioned this to my doctor when I last visited and we saw the same thing with multiple measurements using a more sophisticated machine at the clinic.
I check my BP a lot for reasons, too, and this matches my experience as well. A few tips to reduce the variance (but it'll never go away):
* the position of both of your arms, and the angle your elbows are bent at matters. make sure the edge of your desk isn't pushing into your forearm.
* same goes for both of your legs. make sure you're sitting with legs uncrossed, relaxed angles, make sure the edge of your chair isn't pushing into the back of your thighs excessively.
* control your posture. slouching seems to have an effect too, but I'm unclear on the mechanism.
Anecdotally, the nurses at the doctor's offices I go to rarely (~never) bother to control for any of this other than telling me not to cross my legs. So while I fully believe white coat hypertension is a thing, I also think there's a lot of poor control of variables here. If you look up the AMA/AHA guidelines for blood pressure, the ranges they offer are predicated on some very prescriptive protocols for measurement.. which I virtually never see adhered to.
Is this a digital reader? My readings were always high both at the doctors and at home. Bought an old school analog cuff and stethoscope, now it’s like 120/70 almost every time.
It was apparently the lack of control and the whizzing motor. Having direct fine control over the pump is downright relaxing in comparison.
> I can put the cuff on my arm and sit at my desk for 20 minutes to be nicely rested and calm, and then take 5 different measurements with a few minutes between each one. ...
Doesn't it take more than a few minutes for one's circulation to return to normal after a BP measurement?
I got it when an eye clinic wanted to do my blood pressure (which is kind of weird in itself), which they did shortly after the glaucoma test which never works for me because I have an overdeveloped blink reflex and it turned up high.
It really amazes me how people whose job it is to take blood pressure don’t recognize the stress situations that people are in might result in abnormal readings.
> It really amazes me how people whose job it is to take blood pressure don’t recognize the stress situations that people are in might result in abnormal readings.
I've had a doctor measure my heart rate right after remarking on my "elevated HR during intake" (I walked there and I was running late), after telling me some slightly distressing news on what was already a very stressful day, knowing about my white coat anxiety, and right around the time of peak effects of stimulant medication which they also know about. Wouldn't you believe it, my heart rate was high (110)!
Then I got a front seat to watch their confirmation bias kick in, and have them make some really faulty assumptions based on this bad information. They were quite sure that I had developed a certain medical condition but the lab tests quickly proved them completely wrong.
I had an astoundingly high blood pressure reading last year when I tore my bicep the day after the election and dragged myself to the ER. The staff freaked. I was like, "I had a normal reading two weeks ago in my doctor's office and I'm really in an awful lot of pain and worry right now". But it still earned me the "oh shit hypertension you must do extra pre-surgical clearance".
It was kind of interesting watching my blood pressure tick down back to normal over the course of about 4 days though.
I guess the lesson is don't rip your bicep off, it will transiently elevate your BP. ;)
Beyond any clinical reason that your eye clinic might want to know your blood pressure (your vascular system is pretty important to your vision) - they may have been incentivized by the CMS to track blood pressure via the MIPS program which ties provider payments to specific documentation and screening measures.
AKA - the government might pay your eye clinic more if they screen you for high blood pressure. (Among other things).
Taking a BP in person during a medical consultation, while stressed out because I know I have a very limited amount of time to do a verbal data dump of a large amount of information in a good way never fails to give me extreme values.
It also never helps to explain that this has happened many times before and that I have a reasonably high quality BP measurement device at home (most recently the exact same Braun model) that gives me decent measurements when I'm at rest/relaxed. They just look at me like I'm an idiot. How could a civilian conceivably perform such a complex measurement?
Then I get the predictably insanely high measurement and they look satisfied. Gotcha!
Yep. My heart-rate is normally pretty low (45bpm or so, 55 if I'm walking around), and my BP is on the high end of normal, but white-coat syndrome reliably brings it up to a fluttery 110bpm and 155/125 pressure. I was just in a wreck last week, so I got to see the before-and-after measurements from a few medical providers, and it's shocking how the act of talking to a new person in an unfamiliar environment spikes my BP even when I "know" that it's fine, and also how quickly it falls back down to normal.
If you've got one of these machines at home it seems you really want to run it a couple of times. Often I find the first reading I get is quite alarming but then if I sit for a bit longer and run it another 1-2 times I get a very normal reading.
A big reason people's first measurement is often high is the positioning while putting the cuff on, adjusting to the pose, and then settling in place, a process which can spike pressure. Ideally you prepare yourself (cuff on, in place with your arms at rest, etc) and sit there and relax for five minutes before you take the first measurement, making it legitimately reflective of the at-rest state.
This often isn't possible, but it's a reason doctor visits usually drop you in a room by yourself with a machine that will take multiple measurements with a lengthy delay between. Though that's where the white coat thing comes into play and people are stressed about a doctor coming in, etc.
I might be misunderstanding, but this seems very wrong. your method on the first entry:
10011011010100101100001101010000
year = 101 00 (20? year should be 25 or 2025)
month = 1001 (9, but month is Nov)
day = 1001 (9, not enough bits to encode large days)
hour = 01 1 or 01 11 (3, 7, or 14 reversed, should be 11)
no minute field
I’ve been wearing a heart rate monitor for a while, and after analysing the data, I noticed something interesting: my heart rate spikes in only two situations, when I’m driving, and when I’m talking to my wife.
Oura often interprets gym sessions as stress, and some people on social media argue that it's correct, that you can be stressed at the gym instead of simply listening to their bodies.
Last week, I met an experienced top swimmer who doesn't use any devices, he just listens to his own heartbeat.
So I did something similar (well less cool), but as old Software devs start finding our bodies don’t work as well after a while we will see more and more of this sort of “taking control”
I did wonder but I keep downloading to the app to “see how I am doing”
Throughout the day so I would not be easily able to sniff it - and I assumed that as a highly engineered professional “medical device” it would of course be encrypted with unbreakable … oh it’s probably base64
FWIW the latest version of the app does export the previous day's averages to Apple Health (only when you open the app, mind, which can make it look like there's missing data.) I use BPExtract to read the PDF and export every reading to Apple Health but I'll definitely be giving your stuff a go as well (because automation >> manual every time.)
So this when they became a new brand (?) called Hilo.
I am not following their business but presume something is up.
Of course the format change seems to have broken the extractor so the next free lunchtime(s) I get will fix it. I assumed no one else uses it so was not in a hurry - any feedback gratefully recvd :-)
I am not even that old, but already seeing need to take some things into my own hands. I find going to a GP is more or less just a semaphore for specialists, and those specialists have wait times measured in months to years. I would be insane to just do nothing for that timeframe.
Although I think you need to be quite critical to have such a mindset, and assume you are wrong rather than right.
> I tried feeding a lot of this into various Als (Kagi gives you access to a few with a nice interface) and I found that they mostly were stupid in ways that made me think
"stupid in ways that made me think" is (IMO) a really good summary of how AI is useful, as well as its pitfalls.
This would imply most LLMs on the market are decent rubber ducks. Search engines unrelated. Kagi isn't changing too much (I think) besides adding additional context from search results to your requests for XYZ LLM.
You put on a cuff (supplied) to calibrate it, and then you wear a little gadget on your wrist. It takes frequent measurements throughout the day and night when it senses you aren't moving. It then syncs with your phone to store the results. Its a little pricey, but seems to work well[1]. And it avoids the faff of a pressure cuff[2] and 'white coat hypertension'.
[1] Apart from the option on the app to do a reading from your finger using your phone camera - which gives wildly different results to a cuff.
[2] You need to recalibrate it every month or so using the cuff.
> The Hilo Band uses photoplethysmography (PPG)—an optical sensor technology that collects data from your wrist. This data is sent securely through the Hilo App to Hilo’s cloud server, where advanced algorithms estimate your blood pressure using Pulse Wave Analysis (PWA), which looks beyond just the rate of your pulse, and examines the unique form of each heartbeat’s pressure through your blood vessels.
I doubt that it’s comparable to real blood pressure monitor.
Readit News
I worked next to another engineer in the early eighties who was working on a different project. At tea break time we got into the habit of sitting opposite each other and explaining the problems we were having. Neither of us put much, if any, effort into trying to solve the other's problems because it often turned out that simply attempting to formulate the problem in a way that would make sense to the other party was enough to reveal the answer or at least a promising idea to try.
I also like to do this when debugging. I'll just write out whatever my current hypotheses are about what the bug is and I adjust as I go, as I disprove or make them more precise. I would usually just do it right on the ticket. It both helps because it provides the "rubber duck" and in the end I just submit the comment and I have a root cause analysis right there.
Unfortunately it's quite rare to see this from anyone and asking them to provide a root cause analysis is usually fraught with peril. I don't quite get why. It's been working awesomely.
And like OP says, AI hasn't made it better in the sense that not everyone can use AI (or cow-orkers) in that way. They just believe what the AI tells them and they literally won't reverse engineer that protocol in 24h ;)
> A few times I thought they had "cracked the case" but actually they just made me waste time.
Despite trying various models and services for years now, I've come to the conclusion that the current iteration of AI services hurts more than it helps. I spend more time prompting, reading pages of babble, and trying to find signal in the noise that I would just chilling out and thinking through the problem.
The technology is neat but not productive.
Doctor ends up telling him that he was hoping it was just the attractive nurse, but no. He'd be starting medication today.
Blood pressure is highly variable even moment by moment. A single data point at your yearly visit is basically meaningless. Even your posture can shift it. Legs in a different position? Too much tension in one arm over the other? Seated awkward? Have gas? Unfortunately, it's also very difficult to get a 24 hour monitor unless you have another suspected condition (such as kidney problems).
This prompted me to do some research, and it turns out there's very strict guidelines for how to 'accurately' measure blood pressure, almost none of which were followed by most specialists. There's a laundry list of things that can each increase your blood pressure by 10-15mmHg (eg. drinking water, sitting cross legged, recent physical exercise, not sitting with the right posture, not sitting for at least 15 minutes).
Given this lackadaisical approach to measurement, I seriously question the diagnostic validity of blood pressure tests 'in the wild'.
I have to take daily BP measurements during titration for ADHD medication. (Using an _A&D UA-611 Plus_ machine at home.)
I can put the cuff on my arm and sit at my desk for 20 minutes to be nicely rested and calm, and then take 5 different measurements with a few minutes between each one. They'll vary quite wildly (anything from, say, 115/75 to 135/90) despite not moving between measurements or having any reason to be more or less agitated. I generally just ignore the low/high outliers and average the others. There's no pattern either, sometimes the outliers are first, sometimes last.
Also it's not just the monitor I have at home, the same is true of a probably more trustworthy machine in a clinical setting. I mentioned this to my doctor when I last visited and we saw the same thing with multiple measurements using a more sophisticated machine at the clinic.
* the position of both of your arms, and the angle your elbows are bent at matters. make sure the edge of your desk isn't pushing into your forearm.
* same goes for both of your legs. make sure you're sitting with legs uncrossed, relaxed angles, make sure the edge of your chair isn't pushing into the back of your thighs excessively.
* control your posture. slouching seems to have an effect too, but I'm unclear on the mechanism.
Anecdotally, the nurses at the doctor's offices I go to rarely (~never) bother to control for any of this other than telling me not to cross my legs. So while I fully believe white coat hypertension is a thing, I also think there's a lot of poor control of variables here. If you look up the AMA/AHA guidelines for blood pressure, the ranges they offer are predicated on some very prescriptive protocols for measurement.. which I virtually never see adhered to.
It was apparently the lack of control and the whizzing motor. Having direct fine control over the pump is downright relaxing in comparison.
Doesn't it take more than a few minutes for one's circulation to return to normal after a BP measurement?
It took me a minute to understand (and neither of us think this is 100% true) but it's both funny and a good point.
It really amazes me how people whose job it is to take blood pressure don’t recognize the stress situations that people are in might result in abnormal readings.
I've had a doctor measure my heart rate right after remarking on my "elevated HR during intake" (I walked there and I was running late), after telling me some slightly distressing news on what was already a very stressful day, knowing about my white coat anxiety, and right around the time of peak effects of stimulant medication which they also know about. Wouldn't you believe it, my heart rate was high (110)!
Then I got a front seat to watch their confirmation bias kick in, and have them make some really faulty assumptions based on this bad information. They were quite sure that I had developed a certain medical condition but the lab tests quickly proved them completely wrong.
It was kind of interesting watching my blood pressure tick down back to normal over the course of about 4 days though.
I guess the lesson is don't rip your bicep off, it will transiently elevate your BP. ;)
AKA - the government might pay your eye clinic more if they screen you for high blood pressure. (Among other things).
Deleted Comment
https://my.clevelandclinic.org/health/diseases/25100-hyperte...
It also never helps to explain that this has happened many times before and that I have a reasonably high quality BP measurement device at home (most recently the exact same Braun model) that gives me decent measurements when I'm at rest/relaxed. They just look at me like I'm an idiot. How could a civilian conceivably perform such a complex measurement?
Then I get the predictably insanely high measurement and they look satisfied. Gotcha!
Please bring on the AI doctors.
[0] https://www.garmin.com/en-US/p/716808/ [1] https://www.withings.com/us/en/bpm-connect
This often isn't possible, but it's a reason doctor visits usually drop you in a room by yourself with a machine that will take multiple measurements with a lengthy delay between. Though that's where the white coat thing comes into play and people are stressed about a doctor coming in, etc.
It’s not the cuff position as I used multiple positions, cuffs, and sensors. All 140/90. Plus I feel it.
It’s wild. My BP/HR fluctuates alot outside of clinical, but inside clinical it drops.
0-4 month 4-7 year 8-10 hour 11-15 day 22-23 hour 24-29 flags 30-31 year
I feel healthier after reading this.
Last week, I met an experienced top swimmer who doesn't use any devices, he just listens to his own heartbeat.
https://mikado-aktiia.readthedocs.io/en/latest/
At least you have more than 24 hours to find out!
https://www.nordicsemi.com/Products/Development-tools/nRF-Co...
FWIW the latest version of the app does export the previous day's averages to Apple Health (only when you open the app, mind, which can make it look like there's missing data.) I use BPExtract to read the PDF and export every reading to Apple Health but I'll definitely be giving your stuff a go as well (because automation >> manual every time.)
I am not following their business but presume something is up.
Of course the format change seems to have broken the extractor so the next free lunchtime(s) I get will fix it. I assumed no one else uses it so was not in a hurry - any feedback gratefully recvd :-)
Although I think you need to be quite critical to have such a mindset, and assume you are wrong rather than right.
"stupid in ways that made me think" is (IMO) a really good summary of how AI is useful, as well as its pitfalls.
https://hilo.com
You put on a cuff (supplied) to calibrate it, and then you wear a little gadget on your wrist. It takes frequent measurements throughout the day and night when it senses you aren't moving. It then syncs with your phone to store the results. Its a little pricey, but seems to work well[1]. And it avoids the faff of a pressure cuff[2] and 'white coat hypertension'.
[1] Apart from the option on the app to do a reading from your finger using your phone camera - which gives wildly different results to a cuff.
[2] You need to recalibrate it every month or so using the cuff.
I doubt that it’s comparable to real blood pressure monitor.