Re: Argus usefulness, was Re: Learning from Argus, Re: How to kill open-source project via funding, was Re: Argus correction
On Sun, 21 Dec 2003, Richard D Piper wrote: ... | Specifically, what are the benefits of inter-connecting distributed | systems via email messages instead of https, for example? I think either solution would work. Richard, If either approach is acceptable to you, my preference is to go with https. There are just so many more tools and systems that will be using https (to transport XML). They are both simple methods to transmit patient data, in the absence of a state of the art electronic health record. As I understand it, Argus is also meant to inter-connect diverse electronic health records systems currently in use in Australia? In any case, I anticipate all free EMR systems will be accessible over https anyways (actually, that may already be the case). Unfortunately, although email and https are consumer technologies, PKI/PKC have not reached that status, The current standard appears to be username-password pair for authenticating client-systems and SSL/certificate for servers -- over https encrypted link. Since this is good enough for banking, I suspect it is fine for health information too. Client-side certificate is just too much of a hassle for most people to use. I have not used Argus, but it seems to be trying to address this problem with respect to email. I doubt it has a chance of being successful unless it becomes an opensource/open standard I doubt it. Even if Argus is free/open source, I doubt there will be much interst for it in the U.S.. An advantage of an https approach is that the software (except for the ssl related client) is maintained centrally on a server, ... You bet - and when we tell our patients and administrators: OIO uses the same security as used by banks - they feel very secure. Email evokes thoughts about virus and spam. :-) Best regards, Andrew --- Andrew P. Ho, M.D. OIO: Open Infrastructure for Outcomes www.TxOutcome.Org
Re: Argus usefulness, was Re: Learning from Argus, Re: How to kill open-source project via funding, was Re: Argus correction
On Sun, 2003-12-21 at 21:19, Richard D Piper wrote: An advantage of an https approach is that the software (except for the ssl related client) is maintained centrally on a server, and a remotely installed application (such as argus, which looks complex), does not need to be installed and maintained. Just imagine the problems that could occur with ensuring that version of a particular JVM's on the workstations of 12,000 GPs throughout Australia. Sure, but HTTPS browser-to-web server communication assumes that the practice is always connected to the Internet, which is usually not the case. It is the ubiquitous store-and-forward, asynchronous messaging infrastructure provided by SMTP email which makes it an attractive target. Also, Argus targets automated information exchange between applications, and browsers are not good automation targets, although the HTTPS protocol can certainly be used for automated data exchange (but each application then needs to handle its own store-and-forward buffering, unlike the SMTP infrastructure, which handles it for you). Of course, the Jabber protocol offers both synchronous and asynchronous messaging, and as Horst Herb has said (elsewhere), would be a better choice for healthcare secure messaging (with an encryption layer added). -- Tim C PGP/GnuPG Key 1024D/EAF993D0 available from keyservers everywhere or at http://members.optushome.com.au/tchur/pubkey.asc Key fingerprint = 8C22 BF76 33BA B3B5 1D5B EB37 7891 46A9 EAF9 93D0 signature.asc Description: This is a digitally signed message part
Re: Argus usefulness, was Re: Learning from Argus, Re: How to kill open-source project via funding, was Re: Argus correction
On Sun, 21 Dec 2003 20:56, Andrew Ho wrote: Specifically, what are the benefits of inter-connecting distributed systems via email messages instead of https, for example? Because we *don't* have distributed systems. Interoperability of systems in Australia is very close to zero. All we can hope for at present is to exchange messages in some format most systems will understand (HL7), and the transport of these messages needs to be secured AND *users* (not necessarily systems) need to be authenticated. Horst -- On two occasions I have been asked [by members of Parliament!], 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question. -- Charles Babbage
Re: Argus usefulness, was Re: Learning from Argus, Re: How to kill open-source project via funding, was Re: Argus correction
On Sun, 21 Dec 2003, Tim Churches wrote: ... Sure, but HTTPS browser-to-web server communication assumes that the practice is always connected to the Internet, which is usually not the case. ... Tim, Browser-to-remote web server is not the only possible architecture when using HTTPS. We can also deply Browser-to-local web server-to-remote web server. The local web server can provide store-and-forward function. Best regards, Andrew --- Andrew P. Ho, M.D. OIO: Open Infrastructure for Outcomes www.TxOutcome.Org
Re: Argus usefulness, was Re: Learning from Argus, Re: How to kill open-source project via funding, was Re: Argus correction
On Sun, 21 Dec 2003 21:47, Tim Churches wrote: buffering, unlike the SMTP infrastructure, which handles it for you). Of course, the Jabber protocol offers both synchronous and asynchronous messaging, and as Horst Herb has said (elsewhere), would be a better choice for healthcare secure messaging (with an encryption layer added). Especially since you can easily run XML-RPC (remote procedure calls) via Jabber in the most elegant and fail proof way (including store-and-forward for remote procedure calls!). This allows sort of distributed systems even on the unreliable and slow connetions we have in most of Australia. Horst -- On two occasions I have been asked [by members of Parliament!], 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question. -- Charles Babbage
Re: Argus usefulness, was Re: Learning from Argus, Re: How to kill open-source project via funding, was Re: Argus correction
On Sun, 21 Dec 2003, Horst Herb wrote: On Sun, 21 Dec 2003 20:56, Andrew Ho wrote: Specifically, what are the benefits of inter-connecting distributed systems via email messages instead of https, for example? Because we *don't* have distributed systems. Interoperability of systems in Australia is very close to zero. Horst, I make use of web-based email services and they work quite nicely. Perhaps that could work well in Australia too? Each physician (using web-browser or their EMR) can log-onto a web-site (via https) to retrieve messages meant for them and/or their EMR (via attached HL7 messages). There can be more than one such web-sites - in which case they can swap email messages. These web-base email systems are quite well developed. UCLA happens to use IMP, from the GNU Horde project (http://horde.org/imp/). It can probably be easily modified if Australia needs something special. How is Argus better than this? Best regards, Andrew --- Andrew P. Ho, M.D. OIO: Open Infrastructure for Outcomes www.TxOutcome.Org
Re: Argus usefulness, was Re: Learning from Argus, Re: How to kill open-source project via funding, was Re: Argus correction
On Sun, 2003-12-21 at 21:40, Andrew Ho wrote: The current standard appears to be username-password pair for authenticating client-systems and SSL/certificate for servers -- over https encrypted link. The main problem with simple username/password pairs, HTTPS nothwithstanding, is the need for the client to prove his or her identity to **each** server **every** time the user's password is negotiated or re-negotiated. Such proof needs to be conducted out-of-band to be secure. That kight be acceptable when you are dealing with one server, but typically you would have to deal with tens or hundreds in the course of normal interaction with the health care system. PKI (including the GPG web-of-trust system) is much more scalable. Username/passwords might be OK in a highly centralised system like the UK NHS, but I can't see it being viable as the universal solution to security and authentication here in Australia, where things are much more distributed and organisations more discreet. Since this is good enough for banking, I suspect it is fine for health information too. Client-side certificate is just too much of a hassle for most people to use. Most people deal with only one or two banks at a time, whereas most health practitioners deal with dozens of organisation, and specialist physicians with hundreds (since each GP is effectively a different organisation). That's a big difference between health care and banking. -- Tim C PGP/GnuPG Key 1024D/EAF993D0 available from keyservers everywhere or at http://members.optushome.com.au/tchur/pubkey.asc Key fingerprint = 8C22 BF76 33BA B3B5 1D5B EB37 7891 46A9 EAF9 93D0 signature.asc Description: This is a digitally signed message part
Re: Argus usefulness, was Re: Learning from Argus, Re: How to kill open-source project via funding, was Re: Argus correction
On Sun, 2003-12-21 at 21:49, Andrew Ho wrote: On Sun, 21 Dec 2003, Tim Churches wrote: ... Sure, but HTTPS browser-to-web server communication assumes that the practice is always connected to the Internet, which is usually not the case. ... Tim, Browser-to-remote web server is not the only possible architecture when using HTTPS. We can also deply Browser-to-local web server-to-remote web server. The local web server can provide store-and-forward function. Which web server is that? I bet you are going to tell me that with extra programming to make it do so, any web server can provide store and forward functions for applications which need to communicate with each other. Well, in anticipation, yes, but isn't that what email systems and other purpose-built messaging platforms (like Jabber) are for? -- Tim C PGP/GnuPG Key 1024D/EAF993D0 available from keyservers everywhere or at http://members.optushome.com.au/tchur/pubkey.asc Key fingerprint = 8C22 BF76 33BA B3B5 1D5B EB37 7891 46A9 EAF9 93D0 signature.asc Description: This is a digitally signed message part
Re: Argus usefulness, was Re: Learning from Argus, Re: How to kill open-source project via funding, was Re: Argus correction
On Sun, 21 Dec 2003 22:08, Tim Churches wrote: The main problem with simple username/password pairs, HTTPS nothwithstanding, is the need for the client to prove his or her identity to **each** server **every** time the user's password is negotiated or re-negotiated. Such proof needs to be conducted There is a solution to even this - ssh-agent (see OpenSSH documentation or do a man ssh-agent if you are using a Linux/BSD/MacOSX system). I use it for accessing my surgery server from the hospital via remote X sessions. In theory, each application I start remotely needs authentication. In practice, ssh-agent does this for me. Currently, it times out after 10 minutes or whenever the screen saver kicks in. But it is trivial to configure it so that it works as long as a USB memory stick is plugged in, and you can tie the stick to your belt etc. Horst -- On two occasions I have been asked [by members of Parliament!], 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' I am not able rightly to apprehend the kind of confusion of ideas that could provoke such a question. -- Charles Babbage