Lack of empathy and just not listening to a patient is the death of medicine. Tons of friends with chronic pain of all kinds run into this all the time, especially women. Just being completely dismissed due to lack of a medical degree. I think John Oliver did an episode on something similar in terms of minorities being dismissed.
Look forward to your next developments, mostly happy you found a patch for serious health issues and yeah, I agree totally we will have technology helping to at least monitor the baseline health and detect anomalies in the future - at the same time, doctors will use embedded multidisciplinary diagnosis systems.
Some years ago my father was having terrible a pain attacks in his abdomen. He went to several different specialists, and they gave him all sorts of tests for various things, including an MRI, but they couldn't get to the bottom of it. After more than a year of this, he was getting a regular checkup and mentioned it to the doctor. The doctor said, "Oh. Let me check this." It was a hernia.
Scanning technology is one of those things that I think is advancing way too slowly compared to everything else. Maybe not in research environments, but definitely in real life applications. I hear things about MRIs getting more accurate, smaller and cheaper, but that doesn't change much from the patient's perspective. Imagine MRIs took literally 5 seconds to perform tomorrow. I think there would only be a marginal speed up, as people would still have to go through the process of waiting to see a GP, getting lucky with them believing your concern, getting referred for a scan, travelling to the place to be scanned, checking in, waiting, getting changed, having it done, waiting for a radiologist to give an opinion, then waiting all over again for another appointment to discuss results. Way more annoying if the scan doesn't tell you anything, but if the process was more efficient, then you would have the appropriate information to seek other avenues of data without having to suffer waiting for so long. There's a lot of room for improvement in every one of those steps.
It would be great if our healthcare systems lived up to their ideals, wouldn't it. But in this world, as it is now, we have to be super proactive on all fronts. Appreciate the candid and insightful share as always today Curtis. I'm excited to see the results from this data project. Btw you're playing Satisfactory!?! V.1.0 is quite a step up from the previous versions! A whole new game pretty much
This is on the lines I have been thinking also. I have Horton's Neuralgia among other esoteric conditions and massive issues with getting healthcare professionals to take me seriously. Fortunately the Horton's has been getting slowly better after tooth removal. The tooth was treated in the 80's with poorly executed root canal and possibly, hopefully, the cause of some of the issues. Also, please don't stab yourself on the back with an industrial robot. 😅 I developed a massive fear of needles during a contrast medium injection to my shoulder, but have been combatting it via exposure and steadily getting better at it. Pinching before injection can actually numb the spot.
That’s interesting, thanks for sharing, and don’t worry - I’ll only lance myself with self-retracting disposables (so even if it was something else using the lance, it would just push me away with the plastic housing rather than stab me), but ideally, I’d just like to hold something over my shoulder and get a drop. I have a 3d printer coming, so I’ll be interested to try and play with different designs. I also feel like psychologically hiding the sample area would help. For example, I don’t actually mind ‘pain’ as such, but the context associated with the pain is scary. Hiding the reasons may help. Thanks for the tip about pinching. I also have some trauma-fear from many, many blood tests as a child. I can’t quite get over it, despite trying, every time I have a blood test now, it’s very difficult to handle. It sounds like you understand that well.
@@CurtsStudio My experience is similar, although individual experiences are never the same. I too had many traumatic experiences myself as a child, the worst of which I do not thankfully remember as I was so young, but my parents are still talking about me having nightmares of it for months. I think I understand trauma and fear, but my experience is still only my own. For me facing the fear bit by bit, although very jarring, was the key. Saying aloud that I'm afraid and what I'm afraid of helps quite a bit, and eventually starting to look at the event and trying to form curiosity towards it. For example asking for the children's needle, the one that has the flexible tube, gives me autonomy in the situation more than any other benefit. This might be different for everyone, but my imagination makes scary things bigger if I don't see them, especially true for abstract things, like a fear that I don't quite understand. Oh and one big aspect is that I absolutely need to feel safe about the person holding the needle, if the person feels wrong, or is in a hurry, my amygdala freaks out. Surrounding the event with nice things may also help, possibly by having leftover dopamine and endorphine in the system, or anticipating such. But I genuinely feel that rewarding myself before an event has a bigger positive impact. All this is true also for a dentists appointment. I hope some of my experience translates and is helpful, but I feel like I'm rambling already. Oh, one thing I forgot from my previous message. The tooth that was removed had its roots near or in my nasal cavity, so it was close to the eye and that's why it probably affected the Horton's.
That is a really great idea, the only thing that comes up to my concern is how do you calibrate "normal". I understand if you have a big enough dataset it will slowly correct itself to understand a "normal" pattern. However, what if the data was abnormal from the get-go, like say I already have an undiagnosed random medical condition affecting my metrics right from the very start. This shouldn't be normal however it is normal for my dataset. Also, how should this system correct the expected reaction to treatment while also detecting the abnormality in your dataset? I think this will be one of the biggest obstacles for a comprehensive tracking system.
I don't think these are issues. For the first point, I think it's simply: there's no such thing as a template human, but there are common deviations from an average - it should consider your normal and compare that to average ranges at the same time (maybe have that an optional check). This would be a good thing to do for a user starting to use the system so they can see how far they deviate from a pre-estabilished expectation before it starts caching their own normal over time. As for the user's cached normal, we know what average ranges are healthy for pretty much any metric (especially for blood values). When showing your values, the system can compare with known safe ranges and inform how much of a deviation there is, from which the patient could choose to flag as non-concerning (within a certain range), especially if the cause is a known condition (like a harmless genetic quirk). A patient normal is important to calibrate - like with the breathing issue I mentioned, if my normal had been learnt for my lymphocytopenia, then the abnormal 'normal' range would have been detected immediately, rather than completely missed. For expected reaction to treatment, I don't think the system needs to predict treatment outcomes at all, it just watches 'what is' and identifies anomalies on the cached pattern of that data over time, rather than expecting a specific change caused by a new treatment. The purpose of drugs is largely to stabilize issues, if there is a deviation from the cached norm (that also happens to be closer to known good ranges), then that can still be flagged as an anomaly, for which the patient could selectively flag as non-concerning (as it may be a sign of them getting healthier). If it's a deviation in the negative direction (further away from known healthy averages), then the anomaly can be presented with a higher degree of concern. It would also be sensible when running comparisons in parallel to have the healthy averages be an adapting value (responding to BMI and age). The tracker is not supposed to be a doctor, it doesn't care what medication you are taking, or why. All it does it look for anomalies in data points. From the perspective of user experience, it might be nice to let the patient add their own context to anomalies detected over certain period of time, for things like short-term medications such as antibiotics. But that context should not affect or suppress the anomaly detection, as it may hide a secondary issue, such as a new problem *triggered* by the antibiotics. So to sum it up, I imagine the lowest level of this being pure data-based caching and detection over time, then then next layer is giving that data purpose and comparing that purpose to known healthy ranges for the values, and then the next layer is user context that allows them to selectively flag anomalies as concerning or non-concerning, with the important point being that flagging anomalies as non-concerning *does not* stop them from being tracked.
@@CurtsStudio That makes a lot of sense, however this system will then heavily rely on the user to have a good understanding of their metrix and how your everyday choices could theoretically affect them, and mark them correctly. Since data without or with a wrong context could be as meanless as no data at all. It put a lot of the responsibility on the shoulder of an uneducated individual to correctly decypher the data and react accordingly. But I do believer this is a skill that can be learn through enough exposure, as long as we have the means to track the data people will find a way to understand them. This is honestly one of the rare example that people really try to build something useful out of the AI technology. I wish you the best on the journey ahead.
It would not be mandatory for a user to flag anything. As I said, that would just be an optional user experience concern if *they* wanted context, to functionally stop the system from alerting them on an application level. If anything flags up, that would be a good opportunity seek advice from a healthcare professional, but it also gives the user the chance to make a decision for themselves. Likewise, if they have a concern and nothing flags up, then that helps to reduce anxiety, which to some degree would relieve weight on the healthcare system. Someone could just as well do nothing, look at the anomalies, compare the higher layer context to healthy averages, and present that to someone more informed. The point is it gives more awareness and 'inner sight' to the user. "Since data without or with a wrong context could be as meanless as no data at all." The entire driving force behind the concept, as I explained in the video, is that pattern recognition, [even without any context whatsoever], can still tell you that 'something' is happening, without knowing exactly what it is (which is why context should be a higher layer that *does not* interfere with the anomaly detection process). So I don't see that as an issue either. Finding anomalies in seemingly unrelated connections of data is an unusual way of obtaining information, so it can sometimes be difficult to grasp (because it's not super obvious). Another example of a similar system might be how intelligence services may be able to interpret the population density of a region by measuring how fast people walk on average (or how fast dust accumulates in the corner of a room in an underground station). We could call these data points anything - the point isn't what they are, it's that there's an intrinsic hidden connection between them that affects the value of each. Fundamentally, that's what artificial intelligence is brilliant at, which is why they seem to always find new strategies of playing video games when given the task. If you looked at the data without context, you could still find patterns before assigning any context.
Talking about the health tracking, I'm surprised you aren't wearing a smart watch. Apparently they can be extremely good at keeping track of several biometrics. You didn't mention an end of video emoji so you get this vaguely health related one. Take care of yourself.
I wear an Apple Watch while I sleep and when I exercise (most of the time). I also like that they can do blood oxygen and ECG, even if they’re not as accurate as dedicated devices (still much better than not having the option). I just tend to not wear it while I work because I don’t like it rubbing on my wrist against the table while I use a keyboard. Thanks for watching 🙂
Lack of empathy and just not listening to a patient is the death of medicine. Tons of friends with chronic pain of all kinds run into this all the time, especially women. Just being completely dismissed due to lack of a medical degree. I think John Oliver did an episode on something similar in terms of minorities being dismissed.
Look forward to your next developments, mostly happy you found a patch for serious health issues and yeah, I agree totally we will have technology helping to at least monitor the baseline health and detect anomalies in the future - at the same time, doctors will use embedded multidisciplinary diagnosis systems.
Really interesting discussion
Some years ago my father was having terrible a pain attacks in his abdomen. He went to several different specialists, and they gave him all sorts of tests for various things, including an MRI, but they couldn't get to the bottom of it. After more than a year of this, he was getting a regular checkup and mentioned it to the doctor. The doctor said, "Oh. Let me check this." It was a hernia.
Scanning technology is one of those things that I think is advancing way too slowly compared to everything else. Maybe not in research environments, but definitely in real life applications. I hear things about MRIs getting more accurate, smaller and cheaper, but that doesn't change much from the patient's perspective.
Imagine MRIs took literally 5 seconds to perform tomorrow. I think there would only be a marginal speed up, as people would still have to go through the process of waiting to see a GP, getting lucky with them believing your concern, getting referred for a scan, travelling to the place to be scanned, checking in, waiting, getting changed, having it done, waiting for a radiologist to give an opinion, then waiting all over again for another appointment to discuss results.
Way more annoying if the scan doesn't tell you anything, but if the process was more efficient, then you would have the appropriate information to seek other avenues of data without having to suffer waiting for so long. There's a lot of room for improvement in every one of those steps.
It would be great if our healthcare systems lived up to their ideals, wouldn't it. But in this world, as it is now, we have to be super proactive on all fronts. Appreciate the candid and insightful share as always today Curtis. I'm excited to see the results from this data project. Btw you're playing Satisfactory!?! V.1.0 is quite a step up from the previous versions! A whole new game pretty much
This is on the lines I have been thinking also. I have Horton's Neuralgia among other esoteric conditions and massive issues with getting healthcare professionals to take me seriously. Fortunately the Horton's has been getting slowly better after tooth removal. The tooth was treated in the 80's with poorly executed root canal and possibly, hopefully, the cause of some of the issues. Also, please don't stab yourself on the back with an industrial robot. 😅 I developed a massive fear of needles during a contrast medium injection to my shoulder, but have been combatting it via exposure and steadily getting better at it. Pinching before injection can actually numb the spot.
That’s interesting, thanks for sharing, and don’t worry - I’ll only lance myself with self-retracting disposables (so even if it was something else using the lance, it would just push me away with the plastic housing rather than stab me), but ideally, I’d just like to hold something over my shoulder and get a drop. I have a 3d printer coming, so I’ll be interested to try and play with different designs. I also feel like psychologically hiding the sample area would help. For example, I don’t actually mind ‘pain’ as such, but the context associated with the pain is scary. Hiding the reasons may help. Thanks for the tip about pinching.
I also have some trauma-fear from many, many blood tests as a child. I can’t quite get over it, despite trying, every time I have a blood test now, it’s very difficult to handle. It sounds like you understand that well.
@@CurtsStudio My experience is similar, although individual experiences are never the same. I too had many traumatic experiences myself as a child, the worst of which I do not thankfully remember as I was so young, but my parents are still talking about me having nightmares of it for months. I think I understand trauma and fear, but my experience is still only my own. For me facing the fear bit by bit, although very jarring, was the key. Saying aloud that I'm afraid and what I'm afraid of helps quite a bit, and eventually starting to look at the event and trying to form curiosity towards it. For example asking for the children's needle, the one that has the flexible tube, gives me autonomy in the situation more than any other benefit. This might be different for everyone, but my imagination makes scary things bigger if I don't see them, especially true for abstract things, like a fear that I don't quite understand. Oh and one big aspect is that I absolutely need to feel safe about the person holding the needle, if the person feels wrong, or is in a hurry, my amygdala freaks out. Surrounding the event with nice things may also help, possibly by having leftover dopamine and endorphine in the system, or anticipating such. But I genuinely feel that rewarding myself before an event has a bigger positive impact. All this is true also for a dentists appointment. I hope some of my experience translates and is helpful, but I feel like I'm rambling already. Oh, one thing I forgot from my previous message. The tooth that was removed had its roots near or in my nasal cavity, so it was close to the eye and that's why it probably affected the Horton's.
This sounds fantastic! I wish you all the best in your development journey.
That is a really great idea, the only thing that comes up to my concern is how do you calibrate "normal". I understand if you have a big enough dataset it will slowly correct itself to understand a "normal" pattern. However, what if the data was abnormal from the get-go, like say I already have an undiagnosed random medical condition affecting my metrics right from the very start. This shouldn't be normal however it is normal for my dataset. Also, how should this system correct the expected reaction to treatment while also detecting the abnormality in your dataset? I think this will be one of the biggest obstacles for a comprehensive tracking system.
I don't think these are issues.
For the first point, I think it's simply: there's no such thing as a template human, but there are common deviations from an average - it should consider your normal and compare that to average ranges at the same time (maybe have that an optional check). This would be a good thing to do for a user starting to use the system so they can see how far they deviate from a pre-estabilished expectation before it starts caching their own normal over time.
As for the user's cached normal, we know what average ranges are healthy for pretty much any metric (especially for blood values). When showing your values, the system can compare with known safe ranges and inform how much of a deviation there is, from which the patient could choose to flag as non-concerning (within a certain range), especially if the cause is a known condition (like a harmless genetic quirk).
A patient normal is important to calibrate - like with the breathing issue I mentioned, if my normal had been learnt for my lymphocytopenia, then the abnormal 'normal' range would have been detected immediately, rather than completely missed.
For expected reaction to treatment, I don't think the system needs to predict treatment outcomes at all, it just watches 'what is' and identifies anomalies on the cached pattern of that data over time, rather than expecting a specific change caused by a new treatment.
The purpose of drugs is largely to stabilize issues, if there is a deviation from the cached norm (that also happens to be closer to known good ranges), then that can still be flagged as an anomaly, for which the patient could selectively flag as non-concerning (as it may be a sign of them getting healthier). If it's a deviation in the negative direction (further away from known healthy averages), then the anomaly can be presented with a higher degree of concern.
It would also be sensible when running comparisons in parallel to have the healthy averages be an adapting value (responding to BMI and age). The tracker is not supposed to be a doctor, it doesn't care what medication you are taking, or why. All it does it look for anomalies in data points.
From the perspective of user experience, it might be nice to let the patient add their own context to anomalies detected over certain period of time, for things like short-term medications such as antibiotics. But that context should not affect or suppress the anomaly detection, as it may hide a secondary issue, such as a new problem *triggered* by the antibiotics.
So to sum it up, I imagine the lowest level of this being pure data-based caching and detection over time, then then next layer is giving that data purpose and comparing that purpose to known healthy ranges for the values, and then the next layer is user context that allows them to selectively flag anomalies as concerning or non-concerning, with the important point being that flagging anomalies as non-concerning *does not* stop them from being tracked.
@@CurtsStudio That makes a lot of sense, however this system will then heavily rely on the user to have a good understanding of their metrix and how your everyday choices could theoretically affect them, and mark them correctly. Since data without or with a wrong context could be as meanless as no data at all. It put a lot of the responsibility on the shoulder of an uneducated individual to correctly decypher the data and react accordingly.
But I do believer this is a skill that can be learn through enough exposure, as long as we have the means to track the data people will find a way to understand them. This is honestly one of the rare example that people really try to build something useful out of the AI technology. I wish you the best on the journey ahead.
It would not be mandatory for a user to flag anything. As I said, that would just be an optional user experience concern if *they* wanted context, to functionally stop the system from alerting them on an application level. If anything flags up, that would be a good opportunity seek advice from a healthcare professional, but it also gives the user the chance to make a decision for themselves. Likewise, if they have a concern and nothing flags up, then that helps to reduce anxiety, which to some degree would relieve weight on the healthcare system.
Someone could just as well do nothing, look at the anomalies, compare the higher layer context to healthy averages, and present that to someone more informed. The point is it gives more awareness and 'inner sight' to the user.
"Since data without or with a wrong context could be as meanless as no data at all."
The entire driving force behind the concept, as I explained in the video, is that pattern recognition, [even without any context whatsoever], can still tell you that 'something' is happening, without knowing exactly what it is (which is why context should be a higher layer that *does not* interfere with the anomaly detection process). So I don't see that as an issue either.
Finding anomalies in seemingly unrelated connections of data is an unusual way of obtaining information, so it can sometimes be difficult to grasp (because it's not super obvious). Another example of a similar system might be how intelligence services may be able to interpret the population density of a region by measuring how fast people walk on average (or how fast dust accumulates in the corner of a room in an underground station). We could call these data points anything - the point isn't what they are, it's that there's an intrinsic hidden connection between them that affects the value of each. Fundamentally, that's what artificial intelligence is brilliant at, which is why they seem to always find new strategies of playing video games when given the task. If you looked at the data without context, you could still find patterns before assigning any context.
🤖💉I, for one, welcome our stabby robot overlords.
Talking about the health tracking, I'm surprised you aren't wearing a smart watch. Apparently they can be extremely good at keeping track of several biometrics. You didn't mention an end of video emoji so you get this vaguely health related one. Take care of yourself.
I wear an Apple Watch while I sleep and when I exercise (most of the time). I also like that they can do blood oxygen and ECG, even if they’re not as accurate as dedicated devices (still much better than not having the option).
I just tend to not wear it while I work because I don’t like it rubbing on my wrist against the table while I use a keyboard.
Thanks for watching 🙂