Good and (complex) question. Essentially, the person deciding or purchasing technology in a hospital (an IT administrator) is not the person seeing patients (the doctor). They often work in completely different buildings and no have overlapping experience.
Hospitals already put a bracelet on every in-patient, right? What if you had an RFID chip in the bracelets and scanned at entrances/exits to the ward, to keep track of where the patients are and have been?
why?
I think you are mis-understanding the problem, which is not where patients are but what teams they belong to and what has happened to them overnight
When I first shadowed on a post-op floor in Detroit, I was shocked to see how much information nurses had to copy by hand during a shift change. Even new medicine orders and drug administration schedules were being penciled into the margins of already-crammed pages.
In a lot of ways, nurses have it even worse. They're constantly on the front lines they lose so much time in having to rewrite the same information over and over again.
I have a friend who's been a QA engineer at a couple of hospitals. At one hospital, he noted that the doctors would have their ward meeting half an hour before the nurses would have their ward meeting. One nurse would attend the doctors' meeting, and transfer the information. In order to prevent errors from occurring due to being given second-hand (or the 'transferer' not asking the right question for a patient), he suggested melding the two meetings, which were only half an hour apart. The doctors wouldn't move their meeting "because we consult in the morning", and the nurses wouldn't move theirs because "the doctors meeting happens at a scheduled break time". So errors keep on happening...
He had a few other glorious stories of the amazing politics that goes on in hospitals - sometimes it seems like a bloodsport.
After a 12-hour stay in the ICU a couple of years ago, this doesn't surprise me in the least. I was given a different patient's discharge papers by mistake, and when I got all the records of stay for reference afterwards some of the clinician's reports were utterly illegible. I guess this is why I ended up answering the same medical background questions over and over again during the night.
-
This looks like a great tool How are you going to deal with the inevitable hurdles of HIPPA compliance? (HIPPA = patient privacy laws in the US, for people who don't know).
As a medical technician in a neuro department, I had to take an illegible referral to the neurologist who wrote it (luckily he was in-house). Even he couldn't read his own handwriting: "But I know the patient, and he needs...".
So maybe instead of adding a layer of technology and all of the costs, privacy concerns, reliability concerns, etc., all that's really needed is basic handwriting training, practice, and expectations of legibility for doctors.
I think we do need technology for all the other reasons described in the log post - like the limitations of paper forms, running out of space to add important information, transcription errors and so on. Sure, the digital version could generate problems of its own but many of those are predictable and preventable as they're well known from other systems.
Retraining people to have better handwriting in adulthood is difficult, and arguably it's not a problem of competence as much as conditions - trying to write on a clipboard balanced on your knee or in a rush before seeing the next patient.
It's complicated by the fact that reporting/recordkeeping requirements are not always medically based, but derived from best practices decided by administrators/ regulators/ legislators which are usually put in place with the best of intentions but without full regard for transaction costs. In the medical context, economic friction is often a function of time rather than money; doctors are trying to deliver a certain standard of care to every patient each shift, and administrative work eats away at that time, resulting in an opportunity cost. Filling out a form for each patient doesn't seem like such a bi thing, but the costs of any inefficiency in the form design etc. add up.
Why is this tolerated? It seems like such an absurdly basic problem to fix. Years and hundreds of thousands of dollars of advanced medical training is distilled into a note/prescription that looks like a fucking five year old wrote it. It's pathetic.
A personal anecdote: I went to a doctor recently. He opened my notes, squinted at them, and then asked me what medications he had prescribed me. That was my last visit to him.
It's actually an extremely difficult problem to fix. Anything new that restricts doctors in any way, shape, or form has to come from up high, and it also has to be across the board. Few hospitals have the luxury to turn away doctors, and few areas have enough doctors doing general practice. Make a new regulation, and some doctors will sigh and do it - but other doctors will move on... unless there's nowhere to move to that doesn't also have the new requirement.
When I was a med student in 2009, I did my surgery rotation at a community hospital outside of Boston. Each morning two students were responsible for copying down overnight vital signs from the EHR onto paper. The job took upwards of 45 minutes, and morning rounds were at 6am...
I didn't mind driving through icy Boston roads at five in the morning -- but I did mind the fact that I was doing a job that computers should be able to do far faster and more accurately.
The EHR was a terminal-based system, and I wound up routing traffic through a local proxy, analyzing the logs, and figuring out how the protocols worked. (I'm sure that reading about terminal emulators would have been more effective than reverse engineering them; I can only blame sleep deprivation.) I ultimately built a Python script to drive the terminal based on a list of medical record numbers, recording vital signs and slotting them into pre-formatted progress notes for printing. It even plotted sparklines for fever curves.
It worked. And the chief of surgery loved it. He wrote me a stellar evaluation that prominently mentioned my work with the computer system (I also did good clinical work and all).
Two months later when the medical school dean read my evaluation, I got a very angry phone call. She was horrified that I had "hacked" into the computer system and taken matters into my own hands. I tried my best to explain how the system worked, and why it wasn't a threat. But I didn't really get the message across.
The upshot, though: I realized just how broken healthcare information systems were -- and that I enjoyed working on solutions. After medical school I joined the research faculty at Boston Children's Hospital, where I'm working on open specs, tools, and standards that make it easier to integrate third-party health apps with clinical data and EHRs [1]. It's been an interesting ride -- and I've learned an incredible amount about health interoperability, politics, data, and security along the way.
Do you know the guys working on XTK [1]? They are also working out of Boston Children's Hospital. They have an awesome medical library for webGL based applications. If so, tell them I say hi!
Your story reminds me of my current struggle with a massive medical company in orthopedics and when I learned to never say the word hack again. I was brand new to medical working with WebGL to create a web based bone cutter app for surgeons. Marketing was behind this project idea and had to have it on the Ipad. I used ludei [2] to natively port it to Android and iOS support wasn't completely implemented when I presented to marketing. They demanded it be on the Ipad, I could promise if & when it would be available but suggested a possible 'hack' to get around the Ipad's webGL limitation. They flipped shit on the word hack and went bezerk. I feel your pain.
Somewhat offtopic---do you have an opinion on ludei vs. phonegap, or know someone who does? Working on a js project I'd like to port to mobile, and somewhat bewildered by the choices...
PhoneGap just runs on the built-in web controls of the platform. Ludei reimplements the web control to run faster. Generally you only use the latter for games and graphics intensive apps.
I haven't personally used phonegap. I have had a fried/co-worker who really likes it though. For me personally, ludei solved the problem of getting WebGL to run on the other devices and that's why I used it. I think phonegap is more well known compared to ludei (just my personal feeling) and thus may have a bigger community around it. If it's an application I would say look into phonegap if it's a game look into ludei. Or try both and see which one you prefer.
This is an area that liability is called into play. If the wrong details are printed out for a patient and a bad decision was made based upon them, then you probably know what a hard time you would have had.
It is fantastic to see that you are working to help fix the problems!
> She was horrified that I had "hacked" into the computer system and taken matters into my own hands.
I've never understood this reaction (which I've gotten from elementary school teachers upon showing them a C64 program that printed my name). Probably something to do with insecurity and ignorance.
I'm in the late stages of writing up a PhD on this topic (nursing documentation in a community setting). The major problems are:
1. EHR technology is sold to managers, not clinicians.
2. The leadership of introducing the EHR is fragmented and highly dependent on local conditions such that success in one location does not guarantee success in another location. Likewise with failure.
3. Researchers have a tendency to treat evaluation of health technology as epistemologically equivalent to evaluation of pharmaceutical technology even though a cursory examination of the logic underpinning this assumption clearly demonstrates that this is not the case.
(point 3 makes my job very difficult and political. Fortunately this work[1] and the very few others like it, generally from the same research group lends my work instant legitimacy. Without it, getting my thesis through the committie would I fear be impossible).
my other problem is that I've turned into a moderately successful software developer and the phd is no longer necessary for my career. along with the politics, that creates big motivations problems for me ;)
Hi mate,
I am a Doctor in the startup space for eHR in Australia. I would love the opportunity to further discuss what you have found during your thesis.
If you have any interest at reaching out, my contact details are in my profile.
Cheers,
Rob
My mom who is an NP, (nurse then) used to say 20 years ago that the nurses are the "rememberers" and the doctors, the "thinkers". The most important job of the nurse was to remember what was going on with each patient and get the doctors up to speed as efficiently as possible.
This description of a typical day on the ward is of course ridiculously inefficient to us as digital natives. We deal with information flows every day and can imagine ten different systems that could do this without even opening the door to the hospital.
The big 'but' is that there are so many hidden requirements here. Privacy is a huge concern. Data on patients may not leave the hospital. With the devices we have these days, it is almost impossible to create a near offline system. Almost all devices are connected to "the cloud" in some form or another, making them vulnerable.
The fact that health insurers are grossly incompetent in this aspect does not mean that hospitals should go down to that level.
Another hidden requirement is that this system has to work, always, no matter what happens. Paper notes may not be efficient, or complete. However, you can be pretty sure that while the hospital is still standing, this system works. Making a digital system that is as reliable is a very hard task.
All this will make such a system a huge investment, not only the system itself, but also in training. Balanced against the cost, it is not so evident that this is really an acceptable investment.
The MVP for that system wouldn't be all patient data, it'd be just one metric or something that you could harden the system's other properties against. If you leak one metric, you probably leak them all (and the inverse is probably true). The training costs are amortized because they are dominated by one-time costs of learning the system for one metric, not all metrics.
I dunno about the "always works" claim. I have nothing to do with health care, per se, but I always tell everyone to factor in a base 1%-5% error rate for basic information and data entry that a human beings end up doing manually...
I imagine the number of errors/deaths/kerfuffles due to the opportunity cost of professional time, not to mention transcription and human errors of such systems are not inconsequential...even if they aren't at the forefront of our minds, and even if some of them haven't been measured accurately.
You are never going to take away the human from this equation. Notes are taken over the course of the day, and many of the proposed systems end up taking more time: nurses do their rounds, use pen and paper to take notes while walking around, and at the end of the shift copy the notes from the paper into the system.
This means that you have basically added another data entry level, increasing both time and error rate.
Well I can't comment on the system/app offered by the author of the OP, but obviously the point of designing a system to fix/improve such operations wouldn't just be adding another level of data entry and subsequently multiply risk, but instead to minimize it.
The goal is not to remove humans from the equation, since that is probably both impossible and undesired, but to separate humans and computers into supporting each other in the task at hand by specialising into what they do best: computers for tasks that can be broken down into repetition, replication, automation, speed, validation, and volume, and humans for creation, ambiguity, complexity, context, service delivery and interpretation.
At the moment i'm guessing there's a fair bit of humans doing the jobs computers are good at in hospitals around the world...
It seems like the essentially free work done by interns/medical students is instrumental to these bad systems and procedures continuing.
I wonder what would happen if an intern group got together and said "No. This is a stupid and dangerous way of doing things. We're no longer going to turn up an hour early for a 12-hour shift to copy out notes that should have been taken in a more sensible way to start with."
Interns are typically not fully licensed to practice medicine without supervision, which is the point of a residency program (they take more exams to this effect at the end of the residency). This is why the hospital has enormous bargaining power over them and the hours/wages are markedly different from those of attendings. Unionization of residents is still very rare because the hospital essentially has the power to end the residents' careers for such behavior.
I would love to have a list of tasks for each of my patients on my phone. This would make being an intern far easier. In fact, I'm currently validating and building a solution for this exact problem, which I will talk about below, but first it's important to understand what an intern (at least here in Australia) does during the actual ward round where most of their tasks for the rest of the day are created.
An average intern during a ward round has to do the following things for each patient :
1) Handwrite notes into the patient's bedside notes as the senior doctor takes a history / examines the patient.
2) Look at the patient's vitals chart and medications chart.
3) Handwrite a plan in the patient's notes at the end (this is essentially a list of tasks for the intern to do during the day).
4) Often while the intern is still writing the plan, into the patients notes, the rest of the team is already moving onto the next patient. The intern will hurriedly re-write any tasks from the plan onto their personal printed patient list (takes < 5 seconds) and then quickly go get the notes for the next patient and begin this process again. Also note that often the patients are scattered around multiple wards in the hospital.
Now Listrunner, in their demo video shows a list of tasks for each patient on an iPhone. Awesome!
But where in the ward round does my list of tasks get copied into Listrunner? If I have to manually find the patient in the app and then manually add the tasks to an app it would take minutes, not the <5 seconds it does to rewrite the tasks on a personal list in super shorthand. And no senior doctor is going to wait a couple of minutes for you to write each patients tasks into your phone (this would add 40 minutes to a 20 patient ward round).
I've been thinking about this a lot, and I think a solution using google glass would be super amazing here. I'm currently in the prototyping and validation stage of the project (following Eric Reis' 'build-measure-learn'). Happy to talk to any doctors interested in it.
It works as follows:
1) After you finish writing the patient's plan you take a photo of it with google glass.
2) OCR is performed on the photo, right then an there (hopefully in <= 1 sec) and the OCR is shown to the google glass wearer who can confirm that the OCR is correct.*
3) Those tasks are then synced to the doctor's phone, or for security reasons perhaps a hospital owned phone or tablet.
The advantage of this system is that it doesn't change the current workflow at all. It doesn't affect the speed of the ward round. Thus, faces a lower level of resistance to adoption.
Disadvantage - doctor's are notorious for bad handwriting, thus it will not work for all doctors. It's expensive. However, as google glass (and perhaps other similar tech) gets cheaper this may not be significant.
*Patient labels are already affixed to the top of the page (so OCR can be performed on the label to associate the tasks with the patient). But if the solution became widely used, a simple QR code could be added to patient labels, to make this easier.
I don't think OCR will work that well in this case. Assuming the OCR is able to detect the letters, at least from my experience, they like to draw stuff, write things in uneven 2-column or 3-column format, and make use of shorthand symbols.
> the OCR is shown to the google glass wearer who can confirm that the OCR is correct.
This takes a LOT of time. For a page of patient note full of lab values, one needs to make sure each number is translated correctly. And if something is wrong, how do you expect the wearer to fix the OCR results on the google glass?
Price wise, I would think 1 or 2 google glass would be much cheaper than a real EMR system.
Maybe rather than OCR, just keep the pictures - confirmation then is only required to verify that the pictures are legible. Maybe some effort can be put into triangulating an exact indoor location for each pic, or the order in which patients were visited.
knozi - you bring up a good point, paper is hard to beat in many ways. its very easy to write things down on paper.
Its one of the challenges with Listrunner and we're constantly refining the UX to make it faster to enter tasks. We've looked into OCR but doctors' handwriting is terrible. We have a trial with voice transcription that is promising.
As a techie with a medical doctorate, I've considered these sorts of things too.
The basic problem is that if you want to have records in the computer someone has to input them in a more or less structured manner. This is the issue that is usually left out of EMR discussions. Electronic health records are a trade-off. Everyone realizes there are benefits, but if you ignore the costs when making important decisions, it usually turns out badly.
Because of this, usually what happens is that skilled medical professionals end up saddled with a second data-entry job.
OCR and voice recognition (which you didn't mention, but I think is also a good fit) are both useful tools. However, if you really think about it, the minimum viable product in this space is actually a medical transcriptionist service. Let doctors (and nurses) use the easiest, fastest system they can for recording data (ie paper, or voice), let the DBA's store everything in a massive database for efficient retrieval and make money by providing an efficient reliable method for converting between the two.
i am strongly against qr codes. printed text can be read by both computers and humans. qr codes can only be read by computers, adding an uneeded translation step for the humans.
The article glosses over the fact that wards rounds are typically an exercise that requires vast amount of highly dimensional data. Hidden in the article is the requirement that the information - lab values, vitals, patient location / status, problem list - must be maintained electronically. This by itself is already very hard, especially as paper records never have downtime / network issues and do not talk back when invalid values are written.
Please be assured that the state-of-the-art is not as described. Sufficiently advanced hospitals would have means for healthcare workers to access vitals, lab values, patient location, current medications, current problem list / diagnoses, medical history from previous days and even previous visits, and even medical history from other institutions - all electronically. Staffs enter data into system directly and no transcribing is required. These are available as discrete data i.e. not freetext strings only decipherable by humans.
Also don't forget about electronic prescribing of drugs, glorified vending machines with pockets that only open when there is an order for that particular medication, and barcoded medication administration system.
MD here. I agree that many EMRs can do the things you list. What you have missed is how painful/clunky it is to do so, which is why it's still so often done by hand.
Of course, some EMRs may be good... just none of the 10 or so I have used.
Golly. I thought this was going to be about a hospital ward in sub-saharan Africa, some abandoned corner of Asia, or even a VA hospital here in the US.
I'm glad people care, but for me, these are first world problems compared to drug shortages, inadequate facilities for water and sanitation, missing diagnostics, payroll shortfalls, and so many things.
Sadly, though, these are not problems that programmers can do much to fix.
Oh my god, we still have to write things down on paper?!
Yeah, and so what? Writing things down on paper is a damn tough pnemonic system to defeat.
The problem arises when others have to discern an individual's short hand notes.
So, an array of shitty post-notes is weak and lacks integrity, but similarly, so do disparate plain text files, haphazardly saved in an array of folders with haphazard file names. And yes, you could theoretically grep such a data store for meaningful details, but then again: Have you ever actually tried to parse someone else's randomly sorted plain text notes? It's a brain nullifying experience sifting through someone else's disorganized stream of consciousness.
This is a garbage-in-garbage-out scenario, and coping with a tide of inadvertently injured humanity does not lend itself to well-formed XML and proper SQL grammar. People don't plan on coughing up a lung, or getting shot, or run over. No one plans on syphlis dimentia.
You can bulldoze a landfill of printed circuit boards onto this problem, and still come back and say "boo-hoo, healthcare broken."
It's a hard problem, and there's always going to be a 50% share of elbow grease to pony up on the buy-in. Hospitals are hotels where people disintegrate in the most controlled manner possible.
Feel free to try and automate garbage collection in this environment. You'll assuredly end up with more bloated heap space than a Windows virtual machine running a JVM that emulates a .NET runtime environment.
55 comments
[ 3.4 ms ] story [ 98.0 ms ] thread@jeeshan, I'm assuming you're the author. Why have these practices (archaic maintenance of lists) been maintained?
[1] - HN Guidelines -https://news.ycombinator.com/newsguidelines.html
My bad incorrectly submitting.
He had a few other glorious stories of the amazing politics that goes on in hospitals - sometimes it seems like a bloodsport.
The tech parts of these problems (from personal experience) are pretty mundane and pedestrian--like, summer intern pedestrian.
-
This looks like a great tool How are you going to deal with the inevitable hurdles of HIPPA compliance? (HIPPA = patient privacy laws in the US, for people who don't know).
Something, something razor.
Retraining people to have better handwriting in adulthood is difficult, and arguably it's not a problem of competence as much as conditions - trying to write on a clipboard balanced on your knee or in a rush before seeing the next patient.
It's complicated by the fact that reporting/recordkeeping requirements are not always medically based, but derived from best practices decided by administrators/ regulators/ legislators which are usually put in place with the best of intentions but without full regard for transaction costs. In the medical context, economic friction is often a function of time rather than money; doctors are trying to deliver a certain standard of care to every patient each shift, and administrative work eats away at that time, resulting in an opportunity cost. Filling out a form for each patient doesn't seem like such a bi thing, but the costs of any inefficiency in the form design etc. add up.
A personal anecdote: I went to a doctor recently. He opened my notes, squinted at them, and then asked me what medications he had prescribed me. That was my last visit to him.
I didn't mind driving through icy Boston roads at five in the morning -- but I did mind the fact that I was doing a job that computers should be able to do far faster and more accurately.
The EHR was a terminal-based system, and I wound up routing traffic through a local proxy, analyzing the logs, and figuring out how the protocols worked. (I'm sure that reading about terminal emulators would have been more effective than reverse engineering them; I can only blame sleep deprivation.) I ultimately built a Python script to drive the terminal based on a list of medical record numbers, recording vital signs and slotting them into pre-formatted progress notes for printing. It even plotted sparklines for fever curves.
It worked. And the chief of surgery loved it. He wrote me a stellar evaluation that prominently mentioned my work with the computer system (I also did good clinical work and all).
Two months later when the medical school dean read my evaluation, I got a very angry phone call. She was horrified that I had "hacked" into the computer system and taken matters into my own hands. I tried my best to explain how the system worked, and why it wasn't a threat. But I didn't really get the message across.
The upshot, though: I realized just how broken healthcare information systems were -- and that I enjoyed working on solutions. After medical school I joined the research faculty at Boston Children's Hospital, where I'm working on open specs, tools, and standards that make it easier to integrate third-party health apps with clinical data and EHRs [1]. It's been an interesting ride -- and I've learned an incredible amount about health interoperability, politics, data, and security along the way.
1. http://smartplatforms.org/
Your story reminds me of my current struggle with a massive medical company in orthopedics and when I learned to never say the word hack again. I was brand new to medical working with WebGL to create a web based bone cutter app for surgeons. Marketing was behind this project idea and had to have it on the Ipad. I used ludei [2] to natively port it to Android and iOS support wasn't completely implemented when I presented to marketing. They demanded it be on the Ipad, I could promise if & when it would be available but suggested a possible 'hack' to get around the Ipad's webGL limitation. They flipped shit on the word hack and went bezerk. I feel your pain.
[1] http://www.goxtk.com
[2] http://www.ludei.com
Winslow, your thoughtful comment here is dead, not sure why, best contact admins.
It is fantastic to see that you are working to help fix the problems!
I've never understood this reaction (which I've gotten from elementary school teachers upon showing them a C64 program that printed my name). Probably something to do with insecurity and ignorance.
1. EHR technology is sold to managers, not clinicians.
2. The leadership of introducing the EHR is fragmented and highly dependent on local conditions such that success in one location does not guarantee success in another location. Likewise with failure.
3. Researchers have a tendency to treat evaluation of health technology as epistemologically equivalent to evaluation of pharmaceutical technology even though a cursory examination of the logic underpinning this assumption clearly demonstrates that this is not the case.
(point 3 makes my job very difficult and political. Fortunately this work[1] and the very few others like it, generally from the same research group lends my work instant legitimacy. Without it, getting my thesis through the committie would I fear be impossible).
[1] http://www.ncbi.nlm.nih.gov/pubmed/22188347
For anyone else, the URL is: http://www.listrunnerapp.com/
The big 'but' is that there are so many hidden requirements here. Privacy is a huge concern. Data on patients may not leave the hospital. With the devices we have these days, it is almost impossible to create a near offline system. Almost all devices are connected to "the cloud" in some form or another, making them vulnerable.
The fact that health insurers are grossly incompetent in this aspect does not mean that hospitals should go down to that level.
Another hidden requirement is that this system has to work, always, no matter what happens. Paper notes may not be efficient, or complete. However, you can be pretty sure that while the hospital is still standing, this system works. Making a digital system that is as reliable is a very hard task.
All this will make such a system a huge investment, not only the system itself, but also in training. Balanced against the cost, it is not so evident that this is really an acceptable investment.
I imagine the number of errors/deaths/kerfuffles due to the opportunity cost of professional time, not to mention transcription and human errors of such systems are not inconsequential...even if they aren't at the forefront of our minds, and even if some of them haven't been measured accurately.
This means that you have basically added another data entry level, increasing both time and error rate.
The goal is not to remove humans from the equation, since that is probably both impossible and undesired, but to separate humans and computers into supporting each other in the task at hand by specialising into what they do best: computers for tasks that can be broken down into repetition, replication, automation, speed, validation, and volume, and humans for creation, ambiguity, complexity, context, service delivery and interpretation.
At the moment i'm guessing there's a fair bit of humans doing the jobs computers are good at in hospitals around the world...
I wonder what would happen if an intern group got together and said "No. This is a stupid and dangerous way of doing things. We're no longer going to turn up an hour early for a 12-hour shift to copy out notes that should have been taken in a more sensible way to start with."
An average intern during a ward round has to do the following things for each patient : 1) Handwrite notes into the patient's bedside notes as the senior doctor takes a history / examines the patient. 2) Look at the patient's vitals chart and medications chart. 3) Handwrite a plan in the patient's notes at the end (this is essentially a list of tasks for the intern to do during the day). 4) Often while the intern is still writing the plan, into the patients notes, the rest of the team is already moving onto the next patient. The intern will hurriedly re-write any tasks from the plan onto their personal printed patient list (takes < 5 seconds) and then quickly go get the notes for the next patient and begin this process again. Also note that often the patients are scattered around multiple wards in the hospital.
Now Listrunner, in their demo video shows a list of tasks for each patient on an iPhone. Awesome!
But where in the ward round does my list of tasks get copied into Listrunner? If I have to manually find the patient in the app and then manually add the tasks to an app it would take minutes, not the <5 seconds it does to rewrite the tasks on a personal list in super shorthand. And no senior doctor is going to wait a couple of minutes for you to write each patients tasks into your phone (this would add 40 minutes to a 20 patient ward round).
I've been thinking about this a lot, and I think a solution using google glass would be super amazing here. I'm currently in the prototyping and validation stage of the project (following Eric Reis' 'build-measure-learn'). Happy to talk to any doctors interested in it.
It works as follows:
1) After you finish writing the patient's plan you take a photo of it with google glass. 2) OCR is performed on the photo, right then an there (hopefully in <= 1 sec) and the OCR is shown to the google glass wearer who can confirm that the OCR is correct.* 3) Those tasks are then synced to the doctor's phone, or for security reasons perhaps a hospital owned phone or tablet.
The advantage of this system is that it doesn't change the current workflow at all. It doesn't affect the speed of the ward round. Thus, faces a lower level of resistance to adoption.
Disadvantage - doctor's are notorious for bad handwriting, thus it will not work for all doctors. It's expensive. However, as google glass (and perhaps other similar tech) gets cheaper this may not be significant.
*Patient labels are already affixed to the top of the page (so OCR can be performed on the label to associate the tasks with the patient). But if the solution became widely used, a simple QR code could be added to patient labels, to make this easier.
> the OCR is shown to the google glass wearer who can confirm that the OCR is correct.
This takes a LOT of time. For a page of patient note full of lab values, one needs to make sure each number is translated correctly. And if something is wrong, how do you expect the wearer to fix the OCR results on the google glass?
Price wise, I would think 1 or 2 google glass would be much cheaper than a real EMR system.
Its one of the challenges with Listrunner and we're constantly refining the UX to make it faster to enter tasks. We've looked into OCR but doctors' handwriting is terrible. We have a trial with voice transcription that is promising.
The basic problem is that if you want to have records in the computer someone has to input them in a more or less structured manner. This is the issue that is usually left out of EMR discussions. Electronic health records are a trade-off. Everyone realizes there are benefits, but if you ignore the costs when making important decisions, it usually turns out badly.
Because of this, usually what happens is that skilled medical professionals end up saddled with a second data-entry job.
OCR and voice recognition (which you didn't mention, but I think is also a good fit) are both useful tools. However, if you really think about it, the minimum viable product in this space is actually a medical transcriptionist service. Let doctors (and nurses) use the easiest, fastest system they can for recording data (ie paper, or voice), let the DBA's store everything in a massive database for efficient retrieval and make money by providing an efficient reliable method for converting between the two.
Please be assured that the state-of-the-art is not as described. Sufficiently advanced hospitals would have means for healthcare workers to access vitals, lab values, patient location, current medications, current problem list / diagnoses, medical history from previous days and even previous visits, and even medical history from other institutions - all electronically. Staffs enter data into system directly and no transcribing is required. These are available as discrete data i.e. not freetext strings only decipherable by humans.
Also don't forget about electronic prescribing of drugs, glorified vending machines with pockets that only open when there is an order for that particular medication, and barcoded medication administration system.
Of course, some EMRs may be good... just none of the 10 or so I have used.
I'm glad people care, but for me, these are first world problems compared to drug shortages, inadequate facilities for water and sanitation, missing diagnostics, payroll shortfalls, and so many things.
Sadly, though, these are not problems that programmers can do much to fix.
The problem arises when others have to discern an individual's short hand notes.
So, an array of shitty post-notes is weak and lacks integrity, but similarly, so do disparate plain text files, haphazardly saved in an array of folders with haphazard file names. And yes, you could theoretically grep such a data store for meaningful details, but then again: Have you ever actually tried to parse someone else's randomly sorted plain text notes? It's a brain nullifying experience sifting through someone else's disorganized stream of consciousness.
This is a garbage-in-garbage-out scenario, and coping with a tide of inadvertently injured humanity does not lend itself to well-formed XML and proper SQL grammar. People don't plan on coughing up a lung, or getting shot, or run over. No one plans on syphlis dimentia.
You can bulldoze a landfill of printed circuit boards onto this problem, and still come back and say "boo-hoo, healthcare broken."
It's a hard problem, and there's always going to be a 50% share of elbow grease to pony up on the buy-in. Hospitals are hotels where people disintegrate in the most controlled manner possible.
Feel free to try and automate garbage collection in this environment. You'll assuredly end up with more bloated heap space than a Windows virtual machine running a JVM that emulates a .NET runtime environment.