CPAP software community proposal

[rada]

In this node, I have taken the proposal in the node titled "CPAP Patient Standards Community Proposal" and modified the 'Problem' Statement to respond to Blackspinner's observations about the difficulties of decoding manufacturer data in intermediate files. I will work on providing some scenarios to illustrate what I think could be done.
Claim: Patients can support the collection of data and the development of tools to analyze that data so as to help them treat their own disease and contribute to the body of knowledge about the disease.
Problem: Patients have difficulty achieving compliance with CPAP in part because of the myriad of variables that impact on their treatment and the difficulty of knowing what value of what variable will have what impact on health.
Method to partially solve this problem: A web-enabled discussion forum of some non-profit, relevant entity might be extended so as to support patients, vendors, and health care professionals in developing standards and other tools to address the problem. One view of this virtual organization reflects the traditional product life cycle of requirements, design, implement, and test. In other words: 1) what do people need as regards effective, patient-maintained sleep apnea treatment, 2) how can that be further specified, 3) develop a prototype to demonstrate this better tool or data, and 4) test whether this prototype helps patients. Another view emphasizes standardization and might be seen from the cycle of collecting existing documentation on tools and data, proposing harmonization and updating of that documentation, debating the pros-and-cons of any changes, agreeing on a new standard, publicizing this standard, and seeking compliance with the standard. For instance, the medical profession has an agreed definition of what apnea means that depends on air flow and oxygen saturation; however, patients often have only air flow data and a definition agreed across all parties as to what apnea means when only air flow data is available might be helpful.
Next Steps: Next steps might include improving this proposal, documenting what has already been done to solve this problem, and developing a team of people who are willing to work together across the multitude of tasks and disciplines.

[bearded_two]

In the US and other countries, there are laws, such as the DMCA, concerning the reverse engineering of data, disabling copyright protections, and decryption of data created using proprietary software. Those laws are quite complex and I know enough about the heavy handed application of the laws to know that I have no interest in getting involved in writing software that decrypts any data, including xPAP data.

One example is that a group of three guys in Norway broke the DVD encryption (the program that they released is called DeCSS), only one of the three guys is known. He was arrested and tried in Norway. The the case against him was eventually dropped.
http://en.wikipedia.org/wiki/DeCSS

Another example is a Russian who wrote a program that broke an Adobe encryption. The guy came to the US for a conference and was arrested. He was found not guilty. http://www.freesklyarov.org/

Yes, these two guys were found not guilty or the charge was dropped, but I don't want to risk arrest and bear the cost of defending myself. I would also probably lose my job if I had to defend against criminal charges for violating the DMCA. The DMCA is horrible legislation and the use of the DMCA is reprehensible, but I don't want to risk being arrested or even just being sued by a CPAP manufacturer for decrypting CPAP data.

Another guy was not so lucky. "The Los Angeles jury found 38-year-old Thomas Michael Whitehead guilty on Friday of selling hardware that could access DirecTV satellite broadcasts" "With the six felony convictions, Whitehead faces up to 30 years in federal prison and fines of as much as $2.75 million. Sentencing is scheduled for Jan. 26, 2004."
http://news.cnet.com/2100-1025-5080807.html

Another case in Great Britain: http://www.daledietrich.com/gaming/cate ... nvictions/

[hobbs]

You lost me at "synergistic". Too many Dilbert cartoons I guess.

[rada]

Bearded_two,
Thank you for the 3 links. I followed them and read the articles. In all 3 cases, is the infraction that someone was gaining access to copyrighted information and making that information copyable? Might someone argue that the copyright on my '.rlk' files belongs to me rather than to the manufacturer? Given that the manufacturer has certain patents on its processes, what would I have to do with my '.rlk' files in order to infringe the manufacturer's patent?
I realize there are many laws designed to benefit large organizations with financial interests, which would be the case for ResMed and Philips. So there well may be legal reason to avoid trying to unlock our '.rlk' files -- I wonder whether someone has a pointer to a law or a case that would be closely related to our trying to open our own '.rlk' files for our personal use.
Yours,
Roy
Addendum on my computer and law experience, unrelated though it be: While I know next to nothing about patent law, I have a little experience that might be related. I was a HIPAA consultant and can cite these two books:

• Roy Rada (2003) Health Information Security: HIPAA, Hypermedia Solutions Limited: Liverpool, England and Health Information and Management Systems Society: Chicago, Illinois

• Roy Rada (2005) Privacy and Health, HIPAA, 3rd Edition, HIPAA-IT: Baltimore, MD.

I was indirectly involved in a case about which one could read at http://groups.csail.mit.edu/mac/classes ... crime.html that includes these extracts:
"On June 4, 1993, Neil Woods and Karl Strickland..., became the first two people to be convicted and sentenced for violating the criminal conspiracy provision of the British Computer Misuse Act of 1990. It marked the first time that any successful action had been taken, in the United Kingdom or the Continent, to curb what had ... There are basically two acts relating to computer usage that have been passed by the British government. They are the Data Protection Act of 1984, and the Computer Misuse Act of 1990. The first act generally deals with the actual procurement and use of personal data, while the second act defines the laws, procedures, and penalties surrounding unauthorized entry into computers."
Karl was one of my favorite employees, and I was a character witness for him at his trial in London.
Addendum for hobbs: The dictionary says that synergy means "The interaction of two or more agents or forces so that their combined effect is greater than the sum of their individual effects." Was your point that my writing is too opaque? Would anyone be willing to make my proposal clearer?

[Wulfman]

I am requesting feedback on an idea for a proposal........

......What do you think?

I still haven't been able to follow what your concept is and what you're trying to accomplish.
The software from each manufacturer is different.
The data files and structures are different.......from manufacturer to manufacturer and from machine generation to machine generation.
The manufacturers write different versions of software for each generation of machine they release. (most are backwards compatible)
The software to download and interpret the nightly data from the manufacturer's machines IS already available.....if you know where to look.
The "wheel" has already been invented.
Anybody whose time is worth anything, would only be wasting it.......not to mention the amount of money needed to fight any court battles with the companies who thought their proprietary and copyrighted technology was being tampered with or re-engineered without their written permission.

That's what I think.


Den

[rada]

Den,
The goal is to help patients, such as me, understand their sleep apnea data and adjust their behavior so as to maximize their therapeutic result. The examples are endless but to name a few:

• I adjust the pressures on my APAP but am unclear on whether the results at one setting are better or worse than at another setting.

• Ditto for many other variables, such as the position in which I sleep.

• I may also be unclear on whether I should be minimizing my AHI, maximizing my SpO2, minimizing my ARI, some combination thereof, or something else.

I do not consider practical the alternative of spending every night in a sleep lab and consulting every day with sleep physicians and thus I spend substantial energy on collecting data from sundry tools and trying to interpret the amalgamation of the resultant data. I

• buy a CMS50E and try to relate the SpO2 to the ResScan readings and

• record my sleep with my netbook webcam and try to encode my positional changes and relate those to everything else.

These efforts are typical of many people on this list. For some of these people, including myself, collecting and interpreting this amount of data is challenging. I find this support group very helpful for addressing questions such as 'what is your experience with such a mask and such a maximal pressure'? I am also intrigued by the endless resourcefulness of members of the group at finding new tools and integrating them into their armamentarium, like a reptile heating cable to fashion a home-made heated CPAP hose. However, in relatively short supply seems to be home-made tools to facilitate the collection and interpretation of data. My proposal addresses this latter point.
Part of the preceding discussion at my initiative was trying to get formatting data from manufacturers. I accept based on feedback from blackspinner, bearded_two, and you that that might be infeasible. Forget about that part of it! The issue was how to compare data from different sources. For instance, I have my ResScan charts, my CMS50E charts, and my webcam video. I can compare them all visually one with another and make intuitive interpretations. However, to get more precise I take the numbers from the CMS50E, encode the video into discrete, time series values, and do the same for the ResScan data. That way I can compare the three sources of information more easily. For my CMS50E I wrote some computer code to let me get average SpO2 from arbitrarily specified time periods since I found that useful in testing whether some change in my position in my sleep led to better or worse SpO2. I continue to seek semi-automatic ways to do things and wanted to collaborate with like-minded people.
One issue is standards. How do I want to represent the information from my video -- for instance, do I record positions as prone, supine, and side? For my ResScan charts, what is a useful way for me as a patient to represent that data so as to compare it to my other data -- do I create a table that for each hour notes the average pressure, do I instead keep track of significant changes in the moving average of the pressure, or what?
If I've decided some of the kinds of data that I want to represent and how, then another issue becomes what to do when the data as first presented to me does not come in the format that would seem most useful for further consideration. For instance, I might want to convert the ResScan AHI chart into a sequence of numbers that say at what time I had how much apnea. That's what led to the issue of asking ResMed for the format of its '.rlk' file. But forget that question to ResMed for now. On further reflection, the more prudent approach might be to use existing free software to parse my charts and convert them into numbers -- see for instance http://digitizer.sourceforge.net/. Anything that humans are able to precisely describe doing with data can be automated, and thus looking at the ResScan chart and step-by-step translating that into a sequence of numbers and correlating that with numbers from an oximeter is a process that we can automate.
To recapitulate, I want an understanding of what information I should collect and how I should massage it in order to deal with my apnea and where possible I want examples and tools to support that understanding. What do you want (other than that posts from your colleagues on cpaptalk should be understandable -- and I apologize that my posts might be confusing)?
Roy
Nota bene: In case it is not clear, my big step between this message and my preceding messages is that I

• surrender trying to understand DME, intermediate, data files and

• consider parsing with tools like Engauge Digitizer the graphs the DME device prints.

Another possibility would be to develop tools that compare graphs as images rather than as sequences of numbers. For this insight possibility, I am very grateful to you and others in this discussion.

[Guest]

Roy:

The goal is to help patients, such as me, understand their sleep apnea data and adjust their behavior so as to maximize their therapeutic result.

I commend your efforts in better understanding your therapy as well as your altruistic web-enabled thrust. You obviously have surplus gray matter to apply to whatever intellectual pursuit you fancy. I envy you--I'm still operating in sleep-deprived mode as my therapy slowly improves over time. So bear with me as I attempt to oversimplify your proposal in order to understand it better (and possibly to offer a alternate attack strategy).

In essence you want to be able to:
COLLECT time-phased flow data (apnea events, pressure, etc) from your 'respirator' or flow-generator
SYNCHRONIZE this data with other data collected from other devices (Web Cam, Oximeter, blood pressure, etc.)
ANALYZE these data streams with the goal of adjusting your therapy device (pressure, room temp/humidity, sleep position, sleep schedule, etc).

Let me comment on each of these in reverse order.

ANALYSIS

I may also be unclear on whether I should be minimizing my AHI, maximizing my SpO2, minimizing my ARI, some combination thereof, or something else.

Perhaps you may wish to identify exactly which variable(s) you're trying to optimize first, although this varies by individual. Sleep doctors also disagree on what pressure to prescribe even when presented with sleep lab data. Once you're at an AHI of 5, their job is done (at least in their mind). Also, they seem possessed with the question of 'how you feel', but still rely on subjective quantification methods (Epworth Scale).

SYNCHRONIZATION
The software the sleep labs use seem to be using a standard file format (EDF?) when storing and reviewing time-phased physio data. They combine video data with other stuff too. This might be a good starting point for research utilizing clinical datasets that are already available. Skip the data collection hurdle for the moment and test your therapy algorithms right now.

DATA COLLECTION
This is where we all get hung up on proprietary data formats, conversion, encryption, HIPAA, copyright, etc. Yes, you could wait for Resmed/Respironics to provide what you need. You also could focus on getting legislation/lawsuit together to force them to comply or to adopt open standards.

Alternatively, you could develop a hardware device that could be placed directly in the breathing circuit. I was thinking of some type of flow/pressure sensor with a USB port. Therefore, it would work with any flow-generator. Given this, standards-based open software could easily be developed, effectively circumventing the proprietary legal claims of the vendors. Such a device may already exist within the research community, but has not been productized for the consumer market. Of course, the only bottelneck would be FDA approval. A good entrepreneurial challenge perhaps?

Just my 2cents.

[rada]

Dear OCSleeper,
That's a fascinating idea about putting something in the hose to monitor the pressures so that open-source software and standards could work from there. I have not been active in hardware development and at this late stage in life don't want to start but I will enjoy thinking of that opportunity as a way of understanding what could be.
Good question about the endpoint of therapy as regards personal health. First we get classified by our AHI and then we learn that the AHI may not be crucial. In talking with my sleep specialist on Wednesday after I explained to her my concern that AHI was missing the boat sometimes, she said that I was preaching to the choir as sleep specialists know that AHI is often misleading. That's what led me to SpO2 but she indicates also (as you said) the importance of how one feels. I find 'how I feel' as regards sleep sometimes hard to gauge. My feelings I can better control than my physical health but my physical health is the issue to me now. That led me to ask her about the EEG but she thought I would not be able to benefit from EEG data. I noted how some people on cpaptalk have liked Zeo Personal Sleep Coach and she commented on its advertising campaign. What am I trying to say here? Yes, I agree with you about clarifying the therapeutic target but I do not think it coincides with a single measurement. I do speculative financial investing and there one can relatively easily talk about rate of return on investment. However, investors quickly qualify that with risk. For my sleep I suppose there is a desirable return with high SpO2, low AHI, low arousals, good brain function, good mood, high energy, beneficial cardio and neuro ramifications .... Also there is the trade-off with the effort and cost involved to get these returns ... Finally, as regards health outcomes my proposal was not meant to be about my personal situation but about our situation. So I suppose I would have wanted to explore in the proposed project what do different people find useful. I believe that we fit into many categories and that for each category the beneficial results might be different.
About using data from existing labs -- I agree. Overnight I thought to modify my proposal to working with health care providers and DMEs to come uptodate with their data sets and tools. I am aware of the physionet.org data set and tools. Since I am also an MD, I have been able to go into sleep labs and observe the goings on with data and interpretations. I should get more involved in that. But I also want to relate what the physicians and medical researchers have at their fingertips to what patients have, as I think both groups would benefit from understanding the gaps between the two and the gaps for each group as to what it has versus what it would want for itself.
Not related to your post but also on my mind overnight, allow me to add, that I have found everything on cpaptalk helpful. I commented in my last night's post about the reptile heating cable as a clever thing a patient did. I now use the reptile heating cable. I am trying another mouthguard today based on a cpaptalk recommendation. I have ordered and it should arrive tomorrow the Swift FX nasal pillow based on comments on cpaptalk. I bought the CMS50E based on the advice here. And so on.
OCSleeper, you seem to communicate about some of these matters like I do -- like an academic -- thank you for the feedback and for helping me feel that I was not distancing people too much with my academic approach. My tentative plan however is to leave this thread for a while. My family is calling now and I want to get other things done today and tomorrow.
Roy

[unadog]

In the US and other countries, there are laws, such as the DMCA, concerning the reverse engineering of data, disabling copyright protections, and decryption of data created using proprietary software.

I think a good model in this realm might be the creation and interpretation of "RAW" (proprioetary) data files by pro-sumer and pro cameras, like the Canon 5DII.

In some cases, the manufacturer makes available a "Software Development Kit" (SDK) to facilitate the development of 3rd party software interfaces, like Adobe Lightroom or Photoshop. They actively coopperate with the developers as they release new cameras to help them accurately decript and render the data. The\ manufacturers do so because they see a perceived value to their customers and to their sales from facilitating the conversion of thgeiur files. In some cases, such as Canon's, they also offer their own (free in their case) software conversion tools, so they are in effect "helping their competitors," because it adds overlall value to their product.

In other cases, such as the Phase/Mamyia company, they create and sell their own professional level conversion software. They still assist other developers - Adobe - in converting their RAW files because of markeyt share, etc. But their own software does not support their primary competitors RAW files - Hasselblads - although it does support "lesser" competitors (those not in the medium format digital back market), like Canon and Nikon.

So there are complex examples of companies allowing or encouraging 3rd party development. The question is always going to be how it impacts their own profit, their business model, and how it might enhance (or detract from) the customers perception of their product in the marketplace.

But you really have two seperate questions: access to the data, and the accuracy of algorithms used to interpret the data. I would imagine the latter are proprietary. But sometoimes how you record the data is dependent on the algorythims that use it. So you are treading on a touchy area.

I'll let y'all do the walk on what each manufacturer may or may not support. I'd start with analyzing the market for the existing software tools. Who uses each package, how many people actually buy them, what they try to encourage and prevent (not allowing end-user access to Encore Pro 2.2, for example.)

I can say that prior to the current generation of machines (System One and S9, respectively), they have really done a lame ass job at software developent! What they release reminds me most of product from garage shop developers from around 1995 in ease of use, ease of execution, interface aesthetics, and overall functionality. It is pretty basic!

The S9/ResMed 3.10 screen grabs that I have seen here look significantly better than those from previous software in some ways (breath-by-breath analysis, for example. ) I haven't used the software so I can't really comment on it overall. It might even add enohj value that I woul;d choose an S9 over a System One. But - what percentage of purchasers would say that? And what would get teh DME's - the probable decision maker in a majority of puirchases - to push one machine over another? What would make their relationship to teh customer and job easier so that they pumped more S9's through the pipeline?

Cheers,
Michael

[unadog]

sleep specialists know that AHI is often misleading. That's what led me to SpO2 but she indicates also (as you said) the importance of how one feels. I find 'how I feel' as regards sleep sometimes hard to gauge. My feelings I can better control than my physical health but my physical health is the issue to me now. That led me to ask her about the EEG but she thought I would not be able to benefit from EEG data. I noted how some people on cpaptalk have liked Zeo Personal Sleep Coach and she commented on its advertising campaign.

Maybe **that** is the tool then. A "Workflow Manager and Coach." You get all your data into the system, as it becomes available. CPAP data, Sleep studiy results. Oximeter. Maybe Zeo data summary, or raw? It helps you track and interpret it. Stores it. Helps with longer term management and analysis and trend lines. Include a dail;y sleep diary. How I slept, how tired I was today, 1-10. Memory, Brain fog.

As a byproduct, you keep extending the data collection back "toward" the RAW data. Start with the saved output of the Encore Pro 2.2 reports for example, and "suck" them back in. Or the data from the oximeter. Plus a kind of a sofware version of cpaptalk.com (My mask leaks, then you look at the average leak rate of X mask at X pressure that the folks here put together as a HTML, for example.)

Similar to what Lightroom does in photography. Or a "Cycling "Coach" that I tried to write in an expert system in 1990. I think someone like Chris Charmichal has one now? They might work as "models" for this system. Something that will motivate people to get data into the system. And engage the collective wisdom of the folks here?

Then maybe you get a manufacturer who says "this is a good tool for our clients to use." Maybe you get DME's who say "this can offload some of the endless calls I get about ...." Doctors who say "This is a new way to manage patients. But first, pateients need data capable machines and need access to that data ..." They start to help you, and add additional "prtessure" for the manufacturers to give you access to RAW data. Maybe you create a new model for the management of sleep apnea that certain hospitals and doctors groups adapt?

Just a thought. Can I patent that? Not that I actually want to do any of the actual **work** involved ...

I **think** this is what you are saying above, about parsing data, etc.?

Best,
Michael

[torontoCPAPguy]

It is only a matter of time before the Chinese reverse engineer the whole shebang and offer it on the market for half of what it is being sold for presently AND they will do a better job of it, just like the Japanese and automobiles. Jeez, I cannot possibly imagine ANYONE having driven my Ford Flex before putting it into production. This is our 14th. and 15th. Ford "minivans" (or crossovers if you must) and it is a good vehicle but nothing like the Windstar and Freestar were. We are killing our own economy, much to the benefit of the end user but to the detriment of the economy of course. The end user is benefitting in huge ways.

I recently purchased replacement batteries for our four HP laptops for 1/4 the price that I would have paid to HP. And so on. Why can the Chinese manufacture and deliver a product from Hong Kong or mainland China faster and cheaper than their US distributors? And more efficiently to boot. It is baffling.

My prediction is that most all blowers and accessories are going to be made in China within the next five years (or at least offshore somewhere) and that the equipment will be better, the software will be better and cheaper and so on. I have seen it many many times.

If Respironics/ResMed were to drop the price of their software and make it open source code (or restricted open source) they would wind up with the same bottom line and the end user (that's you and I) would benefit from things like the ability to integrate data, etc.


I'd love to integrate my spO2 levels during the night and my Blood Pressure readings as well. It is NECESSARY information as far as I am concerned and one should not be required to undergo a sleep study in order to see it.

[BlackSpinner]

I can say that prior to the current generation of machines (System One and S9, respectively), they have really done a lame ass job at software developent! What they release reminds me most of product from garage shop developers from around 1995 in ease of use, ease of execution, interface aesthetics, and overall functionality. It is pretty basic!

The S9/ResMed 10 screen grabs that I have seen here look significantly better than those from previous software in some ways (breath-by-breath analysis, for example. ) I haven't used the software so I can't really comment on it overall. It might even add enohj value that I woul;d choose an S9 over a System One. But - what percentage of purchasers would say that? And what would get teh DME's - the probable decision maker in a majority of puirchases - to push one machine over another? What would make their relationship to teh customer and job easier so that they pumped more S9's through the pipeline?

Well this is the part that gets me really really angry, that Resmed has chosen NOT to sell the software to the end users - us. They have already chosen that the DME is their client and that the DME making money off our data is the model they want to follow.

[billbolton]

that Resmed has chosen NOT to sell the software to the end users - us.

Since they DO sell the software to anyone who wants it elsewhere in the world, again you need to look at the North American regulatory environments for healthcare for the source of that particular issue.

Cheers,

Bill

[BlackSpinner]

that Resmed has chosen NOT to sell the software to the end users - us.

Since they DO sell the software to anyone who wants it elsewhere in the world, again you need to look at the North American regulatory environments for healthcare for the source of that particular issue.

Cheers,

Bill

Well I am else where and I can't even get any resmed products shipped to me. I doubt very much I can get it at an affordable price at my DME assuming they knew what I was talking about.

[akhtar]

I am requesting feedback on an idea for a proposal. The broad theme is that home health, patient empowerment, and the web can be combined with open source software and shared data depositories in a synergistic way. One subtheme is that patients can support the collection of data and the development of tools to analyze that data so as to help them treat their own disease and contribute to the body of medical knowledge about the disease. CPAPTALK.com already does a marvelous job at achieving patient empowerment through the web for home health. I am asking about adding a software-related feature.
In the course of trying to deal with my obstructive sleep apnea, I have collected daily for the past two months my ResScan data, my webcam data, my pulse-oximeter data, my annotations about other relevant events, such as mask-type being worn, and have provided daily interpretations of what seems to have caused what and how I should respond differently the next night. I would be happy to share this data, although I expect others are not interested in the detail. However, I do suspect that some might want further tools for handling their data. I want more tools to analyze my data and to help me find relevant relationships in the data. The task is too big for me alone. Could some of us share the effort?
Allow me to illustrate a small part of the task based on my personal current problems. For CMS50E csv files, I can write little Excel programs to help collect trending data from those files, see for example my trivial example in another post. However, what I want to do is semi-automatically link the patterns in the pulse-oximeter data with the patterns in my APAP data. I can manually encode the charts from ResScan into annotations as to the times of clusters of apneas or hyponeas and could write software to relate those patterns to the pulse-oximeter data, but a better approach would seem to be to parse the ResScan rlk files and work on those. However, I don't know how to parse the rlk files and would like to work on that. Someone with a Respironics machine will have data in a different format. Given that we could access some raw, time-series data from some of the vendor products, we could write translation software that could take this data into some agreed formats for further processing with our software that tried to correlate patterns. An example of correlating patterns is to relate changes in SpO2 or pulse to changes in air flow. For my webcam data, I have my own personal coding scheme that I then manually relate to the ResScan and CMS50E data (see for example a sample table relating position and other values), but that coding and relating is extremely time consuming and could be semi-automated.
What do you think?
Roy

I like your proposal. I was thinking of something similar myself, but time is always an issue, and I have several other open-source projects plus work, but this is something I would be interested in contributing to.

I am interested in what you have developed so far, would you be willing to share it?

The data from the Respironics machine I own is imported into a SQL database during the process of reading the smartcard, and since the CME data can also be imported into a database there may be a direct solution for that combination of hardware

Anyhow, I'm just posting on this thread to express interest, and see how many like minded folks there are.