February 2007

customizing health information IN e-age

Views: 187

The mind  needs constant  learning and in medicine,  information is exploding at a rate that is challenging all levels of medical learners. For example in the field of health informatics, the technology involved in keeping patient database needs continual reform to keep pace with the demands of the times. Average patient data, which occupies most of our present day information bases, is often unable to satisfy individual patient needs, in an optimal manner. 

We need an information base that can seamlessly integrate information needs of all categories of certain individual medical learners, with matching solutions offered by other individual medical learners who have already gone through the particular experiences that the other group needs. However, at the very outset the concept of medical learners need to be cleared. In my opinion there is no doctor or patient, but different categories of medical learners. The category I of medical learners comprise of patients and their relatives who are the primary care givers, and category II comprises physicians/health professionals, who are the secondary caregivers. In terms of learning needs, both categories ideally remain life-long medical students. They need to constantly keep learning and updating themselves to keep pace with the vast body of medical information connected by a spidery web that keeps evolving and changing rapidly.

We also need to have a medical learning database where patients and medical students/health care givers regularly key in their narrative logs into a suitable web-interfacing device. Presently PDAs are the closest fit although in the near future it is expected to improve into something wearable with a more efficient input arrangement than the PDA stylus keyboard. This input would simulate a learning neural network with the input channeled to other individual users (with qualitative narrative analysis software to extract themes suggesting information needs from the individual users e-log input). Following this, the network automatically would match each node's (individual user's) information needs through synapses (web based matching, e-mails) and the output could be reiterated several times via a back propagation algorithm to generate an optimal learning solution output. 

Top-down or bottom-up?  

The present system of medical learning, especially during formal training, is more top- down. Future health professionals are very often simply expected to learn and memorize the structure of their chosen field of medicine and then apply it for patient care. However, a complete top-down approach in present day health care is increasingly unable to support health care practice as the volume of information keeps growing by leaps and bounds.

The present day patient satisfaction with health care seems to be at an all time low. One of the important present day problems with this top-down approach is that from the patients' perspective, searching the health professional with the appropriate expertise to tackle their particular individual problems becomes like looking for a needle in a haystack.

One response of the present health care structure to this problem has been an ant like division of labor where health care workers specialize in certain areas so that they can focus on a smaller volume/area of accumulated information and thus offer their expertise in their chosen areas. There is an old adage often used to qualify this approach as, “Knowing more and more about less and less until one has known everything about nothing.”

Interestingly there isn't much historical evidence to suggest that this approach is doing wonders to present day health care. On the contrary present day patient satisfaction with health care seems to be at an all time low. One of the important present day problems with this top-down approach is that from the patients' perspective, searching the health professional with the appropriate expertise to tackle their individual problems becomes like looking for a needle in a haystack.

Also it would be much easier for patients if the large task force of health care professionals kept learning more and more about more and more (instead of less and less) and brought back the good old days of the all-knowing family physician (not necessarily all powerful) who could function as a pillar of strength throughout the whole healing process. Only this time the family physician would have an important check in his omniscient streak; in the form of the well-informed patient. In recent times, the top-down compartmentalized structure of medical education and practice has been challenged by  the evidence-based healthcare movement (before it got compartmentalized itself into a specialty) and the complexity in healthcare movement. Both groups have recognized the need to disseminate learning on a broader basis that bridges the compartmental divide (presently with weak forces) across the specialty structures, in the standard medical model. Bottoms-up learning resurged in medical education in a bigger manner with the formalization of problem-based learning in the 1980s, closely followed by the evidence-based healthcare movement in the 1990s. Evidence-based healthcare is a form of problem-based learning applied to daily health care activities with a purported objective of meeting individual patient needs by using the accumulated and growing epidemiological population based average patient outcome data. At present, it fights a gradually losing battle to establish an absolute unchanging structure of medical truth that is generalized to all users on the basis of statistically averaged health outcomes from controlled trials.

The daily narrative data a patient, health care student/professional generates, reflects both their information needs as well as information contribution (learning points) to the medical knowledge base.

How e-Health can provide the answer

We need a medical learning database that generates narrative informational needs as well as experiences at various levels of medical learning. The underlying theme is that in the daily narrative data a patient, health care student/professional generates and reflects both their information needs as well as information contribution (learning points) to the medical knowledge base. This informational narrative may also address the problem of complex multidimensional needs. Access to life experience information of one patient that finds a match in another patient can itself act as a sort of narrative therapy for that patient.

With improvement in technology, these valuable data of daily processes generated by a given individual can be recorded and documented daily in web based archives. Once in the net, there is technically feasible software already existing that can quickly extract themes from these individual process narratives. Once individual users (the patient, medical student, health professionals) information needs are identified using narrative analysis with available qualitative or other softwares, another software/search engine would automatically collect the matching information available on the web from another individual patient/medical student health professional as well as any related top down theory based information content, and post it appropriately to the site/mail of the individual users concerned (with web links) for detailed reading.

In this way any individual user feeding inputs into the net can receive automatic feedback that can grow as individual users  keep growing and feeding their own daily data. Growing feedback from multiple patients (multidimensional single loop feedback learning) would spur interest and learning in the identified area of need, and can be strengthened by the user returning the feedback with more information/needs to the individuals, who are sending him/her the initial feedback (double loop learning).

Multidimensional Network Learning

The multidimensional learning proposed here is similar to a learning neural network not in physical-mathematical terms but in terms of qualitative analogy as you can see an attempt has been made to depict it pictorially in this slide (Fig 1).

 The individual user (patient, care giver) input is automatically fed to multiple users (nodes) who may synapse with the initial user and other users (via web based matching, email) and finally the resultant learning output shall return to the initial user (by something qualitatively analogous to a back propagation algorithm for example when knowledge is shared with multiple users it keeps getting altered (which perhaps is a hidden layer process among the nodes) and this process may continue to reiterate till there is an optimal solution to the problem.

However, there are several technical problems facing this proposed multidimensional learning method. Firstly, the process is quite time consuming. Recording of daily processes with a PDA at best can be telegraphic information, rather than detailed thought narratives. In the near future however, this may improve with wearable web interfacing devices. Secondly, language will pose a major hindrance, not just in terms of the spoken tongue but even in terms of interpreting symbolic languages like that of mathematics or statistics. They may be undecipherable to the untrained patients.

Interpreting information feedback is another problem area to be pondered upon. Once information needs are identified and fed back to the individual user, interpreting and using them would again be a highly individual exercise and although we would have preferred the information to simulate a structured absolute truth, in the complex real world there may be multiple versions of truth that may vary according to individual user's needs. The evidence-based health care movement at present fights a gradually losing battle to establish an absolute unchanging structure of medical truth that is generalized to all users, on the basis of statistically averaged health outcomes from controlled trials. Critical appraisal is the term for this variety of statistically generalized interpretation but it has demonstrably failed to consistently satisfy the multidimensional needs of the individual patient.

There is also a pressing need for a better web interface, to ensure success of this multidimensional learning method, which in turn can give a positive technological thrust to the world of health informatics. While individual user/authors write, think, and modify their Elogs, a PDA may be grossly inadequate to sustain such an activity. We need a web interface where pages can be turned and information gleaned from multiple pages at one glance; an essential step in multidimensional learning which the present one dimensional scroll panel in PDAs can hardly provide. Lastly patient privacy shall always remain a thorny issue. Privacy can be seen as a way of blocking the progress of learning. As long as we want to remain private we do not want anyone to learn about us. If patients' or health professionals want to learn and meet their information needs they have to share their information with a network, which may compromise the privacy element. 

Even something as personal as a letter is not personal anymore with the advent of email that can be easily forwarded, shared, and not only that, our individual personal letters or E logs (that the present proposal banks on) are being used by search engines to advertise depending on the information needs contained in our letters or E-logs. The future of information age doesn't seem to have much allowance for privacy, and privacy in future will become a direct antagonist of information, unless the future ushers in a new era of information withholding technology.

All said and done allowances have to be made for preserving patient anonymity with due attention to omit particulars that may make it possible to identify users. Speaking from the viewpoint of a developer or vendor of the solution (arising from this proposal) as the individual user continues keying his/her own logs every day, he or she would need to take the necessary precautions and responsibility to protect his/her own as well as the privacy of others he mentions in his/her own write ups.

How individual medical learners utilize their answered information needs would be the beginning of another process in the pursuit of continued bottom-up development in medical learning, which can positively compliment the present pedagogic top-down approach that dominates the current pattern of medical learning.  

References

Biswas R, Always a medical student, Student BMJ(UK) vol 11, Feb. 2003, pg 41. http://www.studentbmj.com/issues/03/02/reviews/41.php
Smith R, What clinical information do doctors need? BMJ 1996; 313:1062-1068
Hodgkin P, Medicine,  postmodernism, and the end of certainty.
BMJ Dec 1996; 313:1568-156

Comments

comments

Click to comment

Leave a Reply

Your email address will not be published.

Latest News

To Top