Friday, May 29, 2009

Cool Technology of the Week

In my recent blog about the Red Flags rule, GreenLeaves commented that biometric checking would help reduce errors by establishing identity and uncovering fraud.

Using biometrics to verify identity seems like a good idea, so I met with Jim Sullivan from BIO-key, a leading provider of biometric solutions.

In the past, I've been reluctant to adopt biometrics because of the expense of buying fingerprint or Iris scanners for each of my 8000 client devices.

However, now that many laptops and hospital ready tablets include embedded fingerprint swipe scanners and that the price of USB fingerprint scanners has dropped significantly, it is realistic to consider biometrics.

BIO-key has developed a next-generation algorithm that reduces the fingerprint to set of calculated unique identifiers. A person’s fingerprint graphic is not the credential; their finger is. BIO-key ensures that only a real finger is being scanned to produce these unique identifiers, making a stolen fingerprint graphic useless to a potential imposter. It's the computed values that are stored when the user's finger is scanned at enrollment, and is later used for comparison with future scans. To me, it's similar to the way NTLM authentication works - there is no need to store or exchange the actual password, it's a mathematical hash of the password that is compared to a stored mathematical hash of the original password. BIO-key allows you to enroll and identify on most of the different fingerprint scanners in the market, allowing an open, heterogeneous fingerprint hardware environment.

There are several interesting ways that biometrics could be used in healthcare:

1. As an alternative authentication method for clinicians instead of having to constantly type a username and password. BIO-key provides a web-enabled fingerprint scanning authentication method that interfaces seamlessly between web applications and an enrollee database or Active Directory. Every authentication, from connecting initially to a secure Wi-Fi hub, to authenticating to Active Directory, to authenticating to web-based or thick client applications, can be done using a finger scan.
2. As a two factor authentication mechanism for secure remote access to sensitive data - instead of a token, you carry your finger with you wherever you go. Note that modern fingerprint scanners include measurement of living tissue, so your finger cannot be stolen and used as an authenticator.
3. As a way to protect patients from identity theft or mis-identification. The first time you register for care, you present your passport and your finger for scanning. On every successive visit, your fingerprint scan is used to verify your identity, without the need to hand-check the paper credentials again.

Some people may think that fingerprints are used to identify criminals and thus be reluctant to use a fingerprint scanner. As noted above, we're not using the fingerprint itself - this is not an FBI comparison to a stored library of fingerprints. Instead, it's comparing the scan of finger to specific computations made on earlier scans of the finger when the patient first registered. Hopefully, this will make patients accept scanning as a positive way to protect their identity instead of a negative "police-like" search of their past.

If you'd like to try this yourself, just get a USB fingerprint scanner or use a laptop with a built in fingerprint swipe reader such as HP, Lenovo, or Dell. Go to http://www.bio-key.com/hitdemo.asp and follow the instructions to download the web client and test the fingerprint enabled applications. Note that it only works in Windows at this time.

A simple way to prevent identity theft and to authenticate web applications using your finger. That's cool!

15 comments:

  1. We have used biometric identification in Oregon to sign death certificates for the last year. One provider didn't want to give a fingerprint (thought that fingerprints are used to identify criminals and thus was reluctant to use a fingerprint scanner) So he uses his toe print to sign - true story!

    ReplyDelete
  2. I had to laugh when I read the comment about the toe, and yes it does work. I don't know what it is about geeky people, and me included here, but why that crazy thought ran through my mind when I got my first tablet pc a couple years ago with the biometric reader on it. I still don't know what came over me, but had to try it.

    After registering my fingers with the software, the next step was the big toe. After the fact I felt a little bit like Maxwell Smart of the 21st century.

    It was fun to try, but like every other civilized geek would do, I use my finger:) It is a good tool when I am mobile tool and very convenient instead of having to use a password, especially with a Tablet PC to boot. Great post!

    ReplyDelete
  3. I implemented a passport/fingerprint physical security solution last year. There are only two drawbacks I found:

    1. Some users just didn't have machine-discernable fingerprints or finger geometries. In my experience it was literally about 1/100th, but still enough to cause an issue.

    2. Wetness and cold-weather were enough to make the biometric machines not register.

    ReplyDelete
  4. On an HP notebook computer, I used a fingerprint reader for about a year. The HP reading software causes a significant delay in every boot. Furthermore, the scanning often took me two or three swipes even after a year of practice. I can type my password faster and more reliably. After I removed the software, I was getting faster boots and logins. When the technology improves, however, I will welcome it again.

    ReplyDelete
  5. This comment has been removed by the author.

    ReplyDelete
  6. HP laptops fingerprint readers are good, but they come with some crappy free software which is what causes the difficulty that Michael describes. We used the BIO-KEY software with that same scanner in the AT&T Wireless stores as our tablet point of sale device and it worked very effectively.

    ReplyDelete
  7. Most fingerprint scanners get fooled through copies of the fingerprint when the person who makes the copy knows what he is doing.

    ReplyDelete
  8. Jim Sullivan here. This last comment is a good example of some of the misconceptions that linger about fingerprint biometrics - and there are many.

    The two misconceptions at work here are that fingerprint technology is static over time, and two, that all fingerprint software platforms are the same. That's like saying, "iPhones can't work securely - a few years ago, my cell phone was cloned.". Do you hear about cloning anymore? The same is true of fingerprint scanners.
    When the low-cost finger scanner industry was first developing, the fingerprint scanner manufacturers focused only on trying to get a decent image from a smaller, low-cost device. Additionally, their software matching algorithms were an afterthought, and had embarrasingly poor accuracy, leading to a very low threshold to cause a false match, as well as an inability to properly match the right person without several retries.

    It is important to understand that BIO-key completely replaces the manufacturer's matching software layer with it's own, causing the same sensor that "just doesn't work without having to re-swipe" to not only work, but be able to work without you having to tell it who you are first (even on a machine that you've never used before). That's because the bad experience people have is usually due to bad software, not a bad sensor, but few people know or understand that. Which leaves the scanner spoofing issue to address

    Once the researchers spoofed the zero-security sensors (as well as MythBusters fooling scanners that already were the joke of the industry with a photocopied fingerprint), the responsible manufacturers (among them Authentec, Futronic, Lumidigm, UPEK and Validity) realized that they needed to incorporate features to assure both liveness and protect the integrity of the data from the scanner to the server doing the match. Using sophisticated anti-spoofing technology built in to the sensor, these modern scanners are not fooled by "someone who knows what they are doing."
    BIO-key has a patented means of securing the entire scanner-to-server pipeline, and it also knows what scanner is being used, allowing the WEB-key administrator to require a certain threshold of security to be met (both on a liveness detection and transport security basis) before allowing a capture and match to proceed. BIO-key operates securely under the premise that the USB port is hacked, the client machine and browser are hacked, the network is compromised, and that a Man-in-the-Middle is inline on every biometric transaction. It even assumes the application server that the surrounding application runs in is hacked. It is designed for online banking authentication, after all.

    So, before you paint the technology that John is using successfully with the colors of you own experience with inferior solutions, I respectfully ask that you actually experience the differences in BIO-key.

    There is a good reason why McKesson, who sat on the sidelines with no biometric offering watching all the problems that Pyxis had with their fingerprint scanners, and who saw the VA shut off Omnicell's one-to-many matching after getting false matches, finally came to market with BIO-key fully embedded. It allows them to beat Pyxis with the fact that, not only does it match better, you don't even need to tell it an ID first, and it scales to staffs exceeding 100's of thousands.

    Similarly, for patient ID, the problem of misidentification is huge, both in terms of Medical ID theft, and as an impact on patient safety. Houston County Hospital has 2500 different women named "Maria Garcia" in their patient database, and saw that they had 3 in the hospital at once on more than one occasion. Name saturation and confusion will only get worse.

    Think about how this technology could address the problem of "Drug Seekers," who hide their identity.

    ReplyDelete
  9. Last night, I had a long conversation regarding biometric and computer aided detection technologies with Jim Sullivan from BIO-key.

    He read my comment on this blog and then decided to reach out to me to explain how much the biometric technology has progressed and especially with their software product.

    Note the term "software product", since they develop the software that uses finger print scanners and not the scanners themselves. BIO-key replaces the vendor-locked matching software that comes with the scanners with their own image processing and enhancement layer, allowing better accuracy, and interoperability among scanners, as well. They do this with the support and endorsement of the scanner hardware vendors, which says something.

    I was impressed by his passion and commitment to his company and its product. I thought he was the founding owner and CEO at first!

    Since I have worked extensively with Single Sign-On (SSO) and CCOW projects and have also worked with Computer Aided Detection (CAD) we were able to communicate without any language barrier whatsoever.

    They have developed an algorithm that resolves issues that the technology suffered from a few years ago when I was experimenting with it.

    CAD has also gone a long way in a couple of years, from almost ludicrous results (5+ ROIs) to an accuracy (<1 ROI) that turns it into a tool instead of being a gadget anymore.

    Their software (BIO-key) runs at a layer above the scanner’s device driver so it can virtually co-exist with other applications that use the fingerprint scanner device.
    I don't know the exact technical requirements but you can find out more at their website: http://www.bio-key.com/fingerprintbiometrics

    I am convinced that this is not an immature technology anymore. They have really pinned down the issues and resolved the nuisances that any innovating technology encounters.

    I can't wait to get my hands on it when I arrive at the office tomorrow morning to test their software on my HP laptop!

    Thanks,

    The EHR Guy

    ReplyDelete
  10. This sounds great! Thanks for the tip!

    ReplyDelete
  11. Coincidentally, I just saw this post on a security blog:

    http://www.schneier.com/blog/archives/2009/05/news_from_the_f.html

    Perhaps we need to be careful about generalizing the utility of biometrics to a diverse patient population...

    ReplyDelete
  12. Jim Sullivan here again. I have to clear up another misconception that Justin's conclusion from reading the "Cancer patient had no fingerprints, was detained at the border" story demonstrates. The misconception is that all fingerprint technologies are the same, so any one system’s failing applies to all such systems. It would be like saying that your iPhone cannot be trusted, because you heard that someone cloned an 8 year old cell phone. Or, that satellite TV doesn’t work, because you heard about a person who couldn’t receive a signal because they lived in the woods, plus don’t forget that satellite dishes are 8 feet in diameter!

    The fact is, this is a *specific* reader technology failing – one that is 8 years old. It should not be surprising that an optical “ridge contour” scanner, as used by Immigration and Customs, would fail to scan someone who has smooth prints – there are no contours, nor ridges, on the surface! To show how far we’ve come since these scanners, those same scanners cost over $500, and are the size of a brick. Modern sensors (now costing less than $3) use RF Technology to look below the skin’s surface, beyond surface wear and tear, to the cell structure that sits below, forming the fingerprint. This individual could have been fingerprinted with the new technology, except that the US-VISIT program is locked to the 8 year old reader technology. This kind of makes John’s case of why he likes BIO-key, since we allow interoperable migration among and between all supported readers, so you wouldn’t be stuck with the old reader technology the way US-VISIT is. It supports the clunky $500 brick and the $3 laptop scanner, and we know which is which, so a more secure application could restrict the use to the most trustworthy devices.

    So why are these misconceptions so readily propagated? Biometrics as a technology is unique in attracting this kind of overly broad, knee-jerk concerns, likely because there is a vocal minority who are scared of the implications of anyone collecting biometric data, who attempt to FUD the technology at any opportunity. They believe, among other things, that the government could use the biometric data to connect your activities across various lifestyle domains. “The government could know that you are the same person who uses a health club as who visited Walt Disney World,” I’ve heard one privacy advocate caution.

    Sounds scary, but let’s actually think about that. Wouldn’t your name and address associated with the health club membership be the same as the name and address on the credit card that you used to buy your Disney tickets, or signed up for the Disney loyalty program with? If so, then isn’t a name and address comparison a much more economical way for a snooping government to connect those dots? It absolutely is, because biometric algorithm matching is much more CPU intensive than comparing text strings. So, unless you are a person who routinely operates under false names and addresses, the prospect of adding a biometric to your health club membership (making it possible to walk in with nothing but your finger), or your Disney Annual Pass (preventing you from sharing your pass), introduces no new risk of big brother watching you.

    My hope is that people will give logical thought to what are the *actual* implications of being able to bind an identifier (Social Security Number, Medical Record Number, etc) to an real person who is the rightful owner of that identifier, preventing, in the process, identity thieves from hijacking it by obtaining some PII from a database somewhere. Before you say, “but a fingerprint could be stolen from a database too,” realize that having possession of biometric data doesn’t convey the identity to the holder – that’s the whole point. Biometric data is measured from a person, not accepted as raw data, and the process by which that takes place is secured against someone injecting data into the pipeline.

    ReplyDelete
  13. well this is interesting as biometrics is something that the world is nowadays looking upto.

    Resveratrol pills

    ReplyDelete
  14. as far as fingerprint readers being compromised by the cold/humidity, finger vein reader technology makes the surface condition of the skin a nonissue. basically if you have blood flowing through your veins the technology works; and if you don't have veins or your blood doesn't flow, it's probably time to re-prioritize your day.

    if you're interested in the technology you can look it up here: http://www.m2sys.com/finger-vein-reader.htm

    ReplyDelete
  15. Miko007, rather than "me too"-ing your way into a discussion about John's success with Bio-key's product, you should probably consider that the extensive annual Unisys consumer survey found that, while over 93% of those surveyed were willing to use fingerprint data to prove identity, just 62% were okay with vein technology. Hmmm.

    See page 17 of the report:
    http://www.unisyssecurityindex.com/resources/reports/US%20Security%20Index%20Oct%2009.pdf .

    Additionally, M2Sys doesn't own the IP around this product, but merely slaps their label on a Hitachi device. One manufacturer - Hitachi - better known for power tools than biometrics, is a customer's only source for this technology - are you serious? If Hitachi hiccups, and, as Fujitsu did this spring, decides to drop manufacturing of their fingerscan devices, everyone who counted on them is hosed. We've seen it happen time and again. Single source proprietary hardware = death.

    I couldn't help but notice that while Bio-key has been killing it in the market with McKesson, AllScripts, Sentillion, LexisNexis, AT&T and other household names standardizing on them, you keep beating your chest about your growing customer base of people like http://www.imcl420.com . Please. ROTFLMAO.

    ReplyDelete