Project Ideas

From Cmsc434_f11
Jump to: navigation, search

Project Ideas

Tom Yeh [1] Kinect

Catherine Plaisant [2] [3]

Here are a few thoughts that I had regarding the class project:

1) How do you display the ability to order left and right radiology images in a CPOE system. I.e. a diagram, a list alternating between left/right body parts, or 2 columns with left and write options for organized by body part. This issue has come up in the past and has lead to patients getting the wrong body part imaged.

2) What is the optimal way to set up a CPOE system where there is a small list of tests that account for 95% of ordering, but you have to have easy access to a 1,000 tests, many of which are very similar sounding?

3) Per our previous discussions, what is the optimal display to limit wrong patient selection errors, i.e. patient photos vs. delayed display of a name, or something else.

4) Building off the lab tracking SHARP-C project. What is the best way to give a a provider a feed of large amounts of information where most results are normal, but there are a few critical results, i.e. signal detection theory.

One possible project might be to have a group of students think about how to embed/visualize patient goals within the context of Nebeker's prototype. I've been thinking a bit about this at a conceptual level and see two possible areas where HCI and information visualization could help:

Todd Johnson <>

1) A tool for capturing and refining patient goals so that they can be mapped onto one or more medical reductionist goals. The idea is that patient goals are likely to be at a variety of levels (or distance from medical reductionist goals). Consider the following goals:

I want to play golf
I want to be able to fly to my daughter's college graduation in California
I don't want to have to get up several times each night to urinate
I want my wrist to stop hurting while I'm using a mouse

These patient goals need to be refined, explored and mapped to medical reductionist goals. In some cases, the goals might suggest non-medical workarounds (e.g., switch from a mouse to track pad). This is a good task for a computer-based collaborative work tool around the more general task of collaborative goal definition and decomposition. There is a social/collaborative element around both refining the goals to reach agreement and understanding between the parties over what would constitute success.

2) Once the patient's high level goals are refined and mapped to medical reductionist goals this mapping needs to be displayed while a clinician interacts with the chart and patient. How can we modify the Nebeker prototype to incorporate the patient goals, and the mapping in way that allows a clinician to balance meeting medical reductionist goals and patient goals? There are interesting issues here involving goals at different levels of abstraction, as well as trade-offs among conflicting goals (e.g., my meds make me too weak to play golf, but if I decrease the meds to improve strength I might be able to play golf, but have a higher chance of having cancer reoccur).

Prescription Decision Support Display (a Possible Class Project, from Victor Gogolak)


Background: There are two aspects of healthcare that are coming together to make prescribing safer:

1) a better understanding, through longitudinal data, of the patient-focused (that is, personalized) aspect of drug reactions, such as age, sex, family history and 2) better means to present physicians the implications of prescribing a drug. The data are voluminous, and are statistically processed to produce rates and counts of reactions under different circumstances. These results are then related to the patient for whom the drug will be prescribed, by advising the doctor what risks to expect for that patient.

Challenge: The results are usually in the form of rates (e.g. incidence of headache per 1,000 Rx) or proportional rates (2.4 times the rate in the general population for ‘headache’), and there are many reactions to consider. Results could list the reaction rates for hundreds of reactions (nausea, headache, stroke, heart attack etc.) and would be enumerated for drugs that could be prescribed, say one of six cholesterol lowering drugs.

Potential Mentor Dean F. Sittig, Ph.D.

University of Texas School of Biomedical Informatics at Houston UT - Memorial Hermann Center for Healthcare Quality & Safety National Center for Cognitive Informatics & Decision Making 6410 Fannin St. UTPB 1100.43, Houston, TX 77030 W: 713-500-7977 F: 713-500-0766

Care Coordination Applications Joris Van Dam <> +1 617 871 4949

Background: In some disease areas, the treatment results that we see in clinical trials is considerably different than the treatment results we see in clinical practice. To use just one staggering example: while in general life expectancy of HIV patients is up to 50 years, there is still an average of 2Mio AIDS related deaths every year – and access and affordability of medication is NOT the problem (says UNAIDS).

This general problem of “translating” the treatment results that we see in clinical research into clinical practice is often referred to as Phase 3 Translational Research – “from bedside to community”, and it is a problem of diffusion, dissemination and implementation of evidence based medicine in clinical practice. Enter medical applications. It is generally acknowledged, and early evidence is already available, that such “eHealth applications” improve treatment results in clinical practice. First and perhaps foremost, it can support patients to better manage their condition and to “adopt the management of their disease in their daily lives”. The Livestrong App is a perfect example. And this is also the main topic of the Interact article (totally different UI, remarkably similar content). Yet secondly, such applications also improve communication and coordination of care between the varied and many caregivers who influence treatment outcome of any individual well being. See e.g. the research around WellDoc – an application for diabetes patients to manage their own conditions. In researching the impact of WellDoc, the patients’ physicians said, that by the information they gathered about their patients through the Welldoc application, for the first time the physicians felt “they were able to manage the disease rather than battle the symptoms”. And thirdly, the reason why companies like Novartis are interested, is that such applications generate data with which we can considerably improve research and development of new and improved treatments (Cf. i2b2. Yet also, if you use these applications both during and after research, you can close the gap between the treatment results that you see during and after clinical research)


So it would seem like a very exciting project to me if you could have a couple of teams designing different applications for the various “actors” involved in the care for a single patient:

-	An application for the patient
-	An application for his physician (specialist)
-	An application for his GP / palliative care physician
-	An application for a caring friend or family member
-	Application for the pharmacy (as pharmacies are starting to play ever larger role in treatment over and above just providing prescription meds)

All centered around the same disease / patient, different applications focused on different “actors” in the overall treatment process, yet sharing some of their data an together all contributing to better outcome for that one patient.

Disease areas where this could be of particular impact: - Rare diseases. At a certain stage in their disease, these patients come home with 1) A devastating diagnosis 2) 24-72 hours of “treatment initiation counseling”, most of which they forgot already because they are still coming to terms with the devastation of the diagnosis 3) a prescription 4) a next doctor’s appointment in 3 months and 5) a perhaps even more devastating feeling of isolation. At that moment and in these situations, such applications can make a tremendous impact in their lives going forward. Cf. Livestrong, cf the Interact article. - Oncology. Particularly in Oncology, patient treatment involve many different “actors” and patients go back and forth between private specialists, general hospital, labs, physiotherapists, ..., ... So the concept or “care coordination” through such applications (as also recently articulated in a white paper that Microsoft wrote about their Health Vault application) is of particular relevance here.

Project partners

I believe it wouldn’t be too difficult to find some folks here who would be interested and available for supervisory phone meetings Most importantly though if you could involve a few patients and possibly their caregivers (perhaps through affiliated School of Medicine) that would probably be of greatest value. Other interested parties could be Ken Mandl of HMS – as ultimately these applications would be implemented on personal health record systems such as the open source Indivo system that Ken is leading. I imagine Microsoft would be equally interested given their efforts with HealthVault, though I haven’t confirmed with them yet.