Slashdot Mirror
Healthcare IT's Achilles' Heel: Sensors
Nerval's Lobster writes "Tech publications and pundits alike have crowed about the benefits we're soon to collectively reap from healthcare analytics. In theory, sensors attached to our bodies (and appliances such as the fridge) will send a stream of health-related data — everything from calorie and footstep counts to blood pressure and sleep activity — to the cloud, which will analyze it for insight; doctors and other healthcare professionals will use that data to tailor treatments or advise changes in behavior and diet. But the sensors still leave a lot to be desired: 'smart bracelets' such as Nike's FuelBand and FitBit can prove poor judges of physical activity, and FitBit's associated app still requires you to manually input records of daily food intake (the FuelBand is also a poor judge of lower-body activity, such as running). FDA-approved ingestible sensors are still being researched, and it'd be hard to convince most people that swallowing one is in their best interests. Despite the hype about data's ability to improve peoples' health, we could be a long way from any sort of meaningful consumer technology that truly makes that happen."
I thought the achilles heel of healthcare IT was the dumbass doctors and nurses who put passwords to everything on sticky notes attached to the monitors?
Film at 11!
I browse on +1 so AC's need not respond, I won't see it.
It's about time people realize this. And it's not just healthcare IT. It's EVERY field. We are SORELY lacking in sensors and data in ALL areas. What are the fucking electrical engineering doing??? Make some fucking sensors!!! The fact that I, as a programmer, still have to rely on GUIs and text as a primary source of input is evidence that more sensors need to be developed. I don't want to have to rely on data entry from fallible humans.
Instead they went for bluetooth and the damn thing says "enter 0000 in the device" and the damned device is buried 3 cm under the subcutaneous layer of the Gluteus maximus. How am I going to do that?
sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
And somewhere an NSA analyst is beating himself frantically thinking about the data collected when insurance companies require* everyone to wear one of these.
*With drastically reduced deductibles!
I disagree with the notion that "it'd be hard to convince people" to ingest medical sensors. Perhaps blood-based sensors would fall prey to that kind of thinking, but those aren't quite possible as of yet. Although a certain number of so-called skeptics will be eager to outline the GPS-tracking possibilities with this kind of technology (which, amusingly enough, could be thwarted by wearing foil), implanted and ingested sensors fill some of the main voids in medical security, namely the current failure to treat chronic illnesses such as diabetes and obesity, as well as the communications breakdown between doctor and patient ("where do you feel the pain?" Is it a stinging or a burning pain?"). Given enough time, the sensors themselves will improve, miniaturize, and drop in cost, and once that happens, medicine itself will change dramatically. It's far too early to call off the parade.
If you need the "cloud" to tell you to step away from that bag of chips, you have issues that technology (short of unpleasant stuff like wiring your mouth shut) can't solve.
The only place they belong is in my colon.
This is an example of over-engineering a solution to the wrong problem. Healthcare is not deterministic, it's stochastic, and we have no direct way of measuring several variables. For example, it's not easy to tell the genetic make up of an individual, gene expression, regulation of such expression or phenotype. We also cannot model environmental impact on the internal biochemical or biomechanical processes. For argument's sake, let's assume that we can do that. How are you going to model how bacteria, viruses, vectors and fomites behave in the real world? Even if you can do all of that, different individuals will have different outcomes in the exact same environment with exact same genetic mark up. If you take two identical twins and subject them to the exact same environment (try it with mice - they reproduce fast and have a short life span), one might develop a disease (e.g. cancers, heart disease, neurological disorders, metabolic diseases) independently of the other. Given that medicine is inherently stochastic, we have learned to reason under uncertainly. We can do a lot of useful stuff in the presence of uncertainty. The only place where you want to measure small amounts of data accurately is for medical imaging of tumor margins, or pharmacodymanics of drugs with very low therapeutic index (the margin between effective and safe dosage). The Achilles Heal is to make the available information well represented and computable, not to spend too much money to measure too little inconsequential data "accurately". What we want help from the technology community are ways to represent clinical, genomic and molecular data correctly and interoperably, and to have tools to discover new knowledge from these data, or to apply existing knowledge to a given patients' data. IAAPAAMI (I am a physician and a medical informatician, with a decent background in computer science).
Replying to myself. The problem with using technology in medicine is not measuring inconsequential structured data "accurately". It is extracting useful information from unstructured data, making sure that the extracted information represents the patient's condition correctly and completely, and that the extracted information is interoperable - for example, you run a patient's document through two different NLP engines - do their output have the same meaning, and can a third machine process those two outputs and tell that they have the same meaning. Most medical records are unstructured free text. You can provide a several gigabytes of exercise or nutrition data for a patient, and a doctor would summarize them in a few pages in a patient's chart that is 200 pages long. The question is, how well can computers evaluate the unstructured data? It's easy to build a new transducer to generate more structured data. It's much harder to build a "transducer" to get accurate and meaningful information out of unstructured data.
Have gnu, will travel.
Yeah, because I find that what I'm really lacking in my life is having a insurance executive following me around every minute of the day nagging me about everything I do.
I can't even trust the government not to snoop on my browser history, and you want me to hand a third party a complete record of my day from the inside out?
Tell a triathlete (or any athlete) that they can get data that can help them perfect their diet or training and they'll try this in a heartbeat.
Can't wait for Clippy 2.0 and all his sensors.
I see you're smoking. Your health insurance has been revoked automatically.
I see you haven't shat in 2 days. I've ordered an enema from Amazon and emailed your office to let them know you'll be out Monday.
I see you're trying to masturbate. Microsoft Bob will read you 50 Shades of Grey to help maintain engorgement of your genitalia.
I think the author is looking at "Healthcare IT" through the wrong lens. The major advances over the next decade are going to come via "big data" (mind the buzzword) and analytics. For example, I was at an EMC presentation the other day and they were making the case for Greenplum (EMC'ized Hadoop / Map Reduce). One of the breakthroughs they were touting was cancer research. There was a woman who went through half a dozen cancer treatments that did not improve her condition. They were able to run models and simulations and ended up finding an off label drug that put the cancer into remission.
Prior to being able to simulate those complex drug interactions, they never would have been able to test an off label drug like that. Now the potential exists to research all sorts of combinations and treatments that would have been computationally impossible five years ago.
Somewhat less cool is the ability to use analytics to track treatment and billing data. The health care system is so inefficient that there is a lot of low hanging fruit there. It only starts showing up at larger scales, once the data sets get bigger. For example, hospitals cannot really correlate the effectiveness of specific treatments for specific conditions versus their peers at other hospitals. Yet with a big enough data set, statements along the lines of, "Last year, 50,000 people were treated for condition X by 50 different hospitals. Of those 50 hospitals, there were 16 different methodologies used. Of those 16, THIS 1 treatment was the most effective, based on these subsequent observations..."
Take a simple use case... the over prescription of antibiotics. With a decent data warehouse and some analytical tools, it will be possible to quickly zero in on cases where antibiotics were inappropriate prescribed. The same thing goes for everything else. "We noticed you are prescribing the brand name medication, when this generic drug is available, and 85% of your peer organizations are using the generic without any loss of efficiency in treatment."
As much as people dislike the government and feel that the Affordable Care Act is a mixed bag, I have a suspicion that when the government gets into really pinching pennies, they are going to drive down the cost of healthcare.
I believe there is an error in this article. In general accelerometers predict upper body motions poorly since they only selectively record the motion of the limb it is attached to. Accelerometers are often calibrated against metabolic systems like VO2 where a regression model is created using raw acceleration data. This type of calibration typically uses locomotion based activities, like running or walking and as a result they don't accurately predict calories for activities that predominantly use upper body motions (like exergames).
M confused, was this article tounge-in-cheek, I can't tell. Doctors aren't looking at fuelbands or fitbit. They are 'connected fitness', not medical sensors... There are studies showing that people using them are 10% more active than others when they have friends also connected and some competitiveness. Sure they're not accurate, but they're indicative.
In theory, sensors attached to our bodies (and appliances such as the fridge) will send a stream of health-related data -- everything from calorie and footstep counts to blood pressure and sleep activity -- to the cloud
No fucking way. My sensors will send my data to my computer in my house where no one else has access to it. Any automated analysis will be done on my computer where I control who can see the results. Any other design is nothing more than putting the band-wagon ahead of patients' interests.
When information is power, privacy is freedom.
It seems to me that really we as a people have some "big data" privacy concerns to work out first. Just like the plugin obd2 monitor for your car that lets the insurance company know you are behaving (no matter what your driving record is), I think people will have a generally negative take on a refrigerator that tattles on them. Also - what's in it for them? There has to be something there that gets them to buy in, and so far I'm not seeing it. People who are healthnuts are already tracking their stats without the technology. The rest don't care (look at their sweat pants).
Initial clinical tests of this technology have been very positive with regards to the effect on healthcare provided.
http://www.xsensor.com/Foresite
So the poor Sensors can be used to black list people
Sure, if they can get the data for free.. But would they be willing to pay, say $10k/year? 10k one time, and 1k/year?
they can get some data now for around $100-200 (the shoes and fitbit type devices)... But that's down in the "buy an expensive toy" kind of territory, like things at Brookstone.
Serious athletes pay thousands for high end bikes, skiis, etc. What about pro athletes?
'Despite the hype about data's ability to improve peoples' health.'
Healthiness, here's how:
Eat real food (mostly vegetables and meat/dairy. Not too much grain. Not too much in all).
Get plenty on outdoor activity.
Drink moderate amounts of alcoholic beverages (about 1.5 - 2 times the "recommended levels").
Don't smoke.
Have an engaging and intellectually challenging hobby.
Brush your teeth.
Do fun things.
Great minds think alike; fools seldom differ.
higher muscle mass tends to put them in the overweight category by BMI
BMI also does not account for body frame size
Poor eating habits & inadequate exercise are the problem. But heaven forbid people are responsible for their own actions.
I want to delete my account but Slashdot doesn't allow it.
For example given the blood pressure monitors one can buy, its just a matter of networking them to a smart phone, likewise blood sugar monitors, heart rate monitors and scales which all exist, its just a networking problem, and as hinted in another post this is where bluetooth might work. If your talking monitoring an elderly person living alone, motion detectors exist, as do water flow detectors, so that for example if there is no motion in a bathroom for 10 hours and no water flow call for help.
The online fridge has been a pipe dream for 20 years and will likely not be widely used for at least 20 more years . If you switch from the pendant model of calling for help to a wrist watch model again most of the tech exists.
Now the dreams of detailed monitoring raise issues, and I agree that I would not want them, and would connect any internet connection that monitored the contents of my fridge to the bit bucket only.
Now that we have Obamacare, why can't HIPAA be crumpled up and tossed?
The whole point of HIPAA was to keep your health info secret from insurance companies, so that they wouldn't deny you coverage or overcharge you for being a sick person. Of course, that was a stupid idea- insurers don't care about the juicy details of why you're sick. They only care about which policyholders are costing money. You obviously can't keep your insurer from finding out about the claims payments they're making on your behalf. And nothing ever seriously prevented insurers from conspiring with each other about whether to blacklist you, even if your doctors were gagged and prevented from revealing your blood type to each other.
The central feature of Obamacare is supposed to be preventing insurers from overcharging or denying coverage to people with preexisting conditions. If that's the case, then WTF is the purpose of HIPAA anymore? Keeping your spouse in the dark about your herpes? Making sure potential bosses can't google your name against "DSM-5" or "virus"? Preventing WalMart greeters from tackling kleptomaniacs? Ensuring food charities won't withhold green bananas from the terminally ill?
Since HIPAA was enacted, people have really gotten in their heads that the privacy of their health information is their most treasured possession. People are even screaming about HIPAA violations on healthcare.gov because it has a radio button for smokers to click on. But if *insurers* can't do anything with this information anymore, then who cares? Should we still be forcing sick people to make extra trips to the hospital so they can pick up their X-rays in person? This info should still be confidential, but I don't know why it deserves its top secret status anymore.
The British NHS serves a population of ~50 million. Kaiser Permanente has about 10 million. Those will be big enough sample sizes to get most of the big data info.
I entirely agree that BMI is a flawed system. It doesn't pretend to be an absolutely foolproof statistic, either. Body fat percentage is more indicative of overall health, of course, but the main issue with using body fat by ratio is that it's much harder to measure accurately. And that's a recurring theme within medicine, especially the types of medicine that focus on improving the overall health of the patient, rather than simply finding a problem to solve. It's a problem that implanted and ingested sensors will solve rather than exacerbate, because they allow for more accurate measurement in an easier, more constant manner.
Eventually insurance companies will make use of these sensors required in order for you to purchase their policies (or at least not pay exorbitant rates), and of course they'll penalize you if the sensors detect you living an unhealthy lifestyle, not following your doctor's instructions, and so on. And now that we have compulsory insurance in this "free country," there'll be no escaping being surveilled and coerced like this.
Aren't you glad you supported Obamacare? Say goodbye to yet more of your freedom.
Liberty in your lifetime
A coworker had to wear an activity bracelet for a few days for a health assessment. He put it on his ankle and tapped his foot or bounced his ankle most of the time. The results showed his 5 foot 11 inch (1.8m) and 285 (129kg) Eric Cartman frame was walking/running +7 miles (11km) per day.
No good deed goes unpunished.
Perhaps if they put a sensor on the heel they would be able to detect this problem.
I'm a good cook. I'm a fantastic eater. - Steven Brust
Our society itself is sick. The problem is, thus far, not worded, but is trying to be masked with technology. Our bodies are evolved from billions of years of Nature's advancements with it's own set of bio-logic. Until we learn to utilize that bio-logic for the good of all, we are simply bullshitting ourselves into thinking that we're innocent, and have no wrong-doings in our diet, physical habits, or our thought habits.
Politics; n. : A religion whereby man is god.
OMG how cutely clueless. Doctors today recommend behavior and diet changes and nobody listens already.
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.
I work in healthcare IT. One of the biggest problems we're facing stems from HIPAA, Hi-Tech and AVA. Due to ACA (Obama are), we receive lower reimbursements from Uncle Sam than the ones we used too (that still didn't cover the cost of providing care). The catch 22 is that thanks to compliance with data security, all of our data has to be hosted in the US and only accessed by persons residing in the US. While it is a bit of job security that my job won't go to India, many of the companies we deal with on a daily basis (HP, IBM, Symantec, Dell, Red Hat, etc) run their support and remote services primarily off-shore. Even though there isn't a direct requirement to keep data in the USA, CMS and the OIG made it very clear they will not pursue litigation across the US border and that it will pursue the last entity to have responsibility in the US.
According to my Nike Fuel Band, I just masturbated two miles.