Wishing for a smart health search

Health care consumers today want to use the Web to find information online about doctors, specialists and care in general. And they want it to be useful.

Unfortunately, in the vast health search space based mostly on ad revenue and keyword densities, consumers often spend hours clicking links into dead ends and wind up with no more knowledge for their trouble.

A semantic Web promises more accurate and meaningful results, yet this technology is in its infancy. And most “trusted” health sites do not yet support semantic searches. Moreover, semantic search requires some knowledge of how to construct a search query as opposed to a simple Google-style search.

To illustrate the limitations of the current search model, here’s what came up when I asked, "What do I do if my head hurts?"


Most searches today, whether health-related or not, are a one-time
game. Search engines do not remember the searcher or his or her
background, history and preferences. Searches for doctors, educational
materials and medical services could especially benefit from this
information, as a lot of information is encoded in the health profiles,
personal health records, history of searches and communications with
other people. Social Network sites have a lot of this information and
potentially could take full advantage of using it to do highly targeted
and personalized searches.

Applying these tools to searches for health care providers could allow
consumers to see a filtered view with resources in their local area,
providers affiliated with certain insurance plans, and providers that
were effective for other consumers in their peer group.

These “filters”
allow the consumer to follow a careful direction through the maze of
health care providers. In addition to great guidance through filtered
health care provider searching, the consumer has a right to see what a
provider has accomplished for previous and current patients. When
consumers can read reviews, rages and raves about a potential health
care provider, they are able to choose who fits their needs more
appropriately, as opposed to wasting their time visiting one doctor
after another just to find the right one.

While Social Networks themselves cannot help with diagnostic-related
searches and answer questions like “What do I do if my head hurts?”,
the peer-to-peer communication is essential and has its own place in
the overall health-education space. Knowledge passing from one consumer
or patient to another may open doors and add insight to treatment
options and ideas that were previously unknown to that consumer. Is
that safe? I it is when the information passed between users and
patients is not medical advice and offers no medical guidance, but is
more of a psychological bond and literacy channel to share personal

The knowledge can also flow the other direction. After all, health care
is an intermingled web of doctors, providers, patients and information.
The information floats freely in such a network, and better tools are
needed to ease the process of tapping into that knowledge. Perhaps a
search engine equipped with the knowledge accumulated in discussion
groups and forums and is semantically linked to the medical
recommendations could share existing wisdom with the searcher,
especially if the search engine knows who the searcher is.

There is definitely a lot of hope for technology to help consumers in the health-search space.

Here’s my wish list for a Smart Search Engine:

Remember who I am.

  1. Filter the information that is relevant to me.
  2. Correct me if my question does not make sense.
  3. Suggest me “You could be also interested in …” material.
  4. Alert me if my clicks and pages have semantically deviated from
    what I’m searching for and let me get back to the starting point.
  5. Remember what I already read.
  6. Learn my social interactions with others, and deliver me the
    wisdom of others, especially if others are qualified or really “close”
    to me and my medical problems.

I really hope to see this engine one day — better sooner than later. I
want to see the intelligent response, and I want this engine to learn
about me, to age with me, and to help me with my health issues when I
really need it.

Dan Kogan is a veteran IT developer who built some matching systems now prevalent in financial institutions. He’s now turned his attention to healthcare and has started Health World Web.

Categories: Uncategorized

Tagged as: ,

4 replies »

  1. Another way of dealing with this problem would be to have human infomediaries ( for example, nurses or librarians in India) help users to find relevant information. This could be done inexpensively ( less than a dollar per query answered) and would help users to find high quality, relevant information. This is using the mechanical turk model to improve the results of search engine – a blend of both the search engine algorithms and human expertise !
    Dr Aniruddha Malpani, MD
    Medical Director
    HELP – Health Education Library for People
    Excelsior Business Center,
    National Insurance Building,
    Ground Floor, Near Excelsior Cinema,
    206, Dr.D.N Road, Mumbai 400001
    Tel. No.:65952393/65952394

  2. Dear Dan,
    Great wish list !
    One simple way of doing this would be to provide every user with a PHR – and the information in this PHR could be used to tailor the search engine results, so it delivers only relevant information.
    Dr Aniruddha Malpani, MD
    Medical Director
    HELP – Health Education Library for People
    Excelsior Business Center,
    National Insurance Building,
    Ground Floor, Near Excelsior Cinema,
    206, Dr.D.N Road, Mumbai 400001
    Tel. No.:65952393/65952394

  3. Max.
    Thank you for your comments. You are raising the valid points. The search engine that we are building at HealthWorldWeb.com deploys various areas of machine learning techniques to filter the topics by the user preferences, not just matching the zip code or tracking user’s visits. Moreover, we take the privacy of our users very seriously, so we neither spy, nor record every click. For a smart engine this is not a requirement.
    The user’s choice is a primary driver behind a good adaptive search system. That’s why we believe that “good search engine” is the one that learns the searcher’s habits and preferences and produces results accordingly. For example, if the user chooses language, or insurance coverage as the primary constraints, they would see different results as compared to those users who selected zip code as the main criterion.
    User generated content is important and the question of trust is the user’s personal choice. As we will deliver the content based on the preferences, your results will be different from other users. There are different techniques and algorithms available to make this possible.
    As for the source of the search being the only requirement, we believe that the best search engine should combine a trusted information source, adaptive personalized filtering, personal preference, solid analytics and smart UI.
    Kudos for the detailed review.
    Dan Kogan

  4. Some good points here that are potentially valuable. However, I caution on a few items to consider:
    1. If you are only offered searches based on your previously automatically saved search records, and/or based on your geographical location, how would one have all of the knowledge available on the topic open to them? For instance, local doctors and healthcare facilities in my area are very limited. I can visit, and actually have visited, all of the healthcare sites in a two city area. I don’t want to be shown primarily local information.
    2. If the search engine remembers and records everything you read automatically, without regard to preference, aren’t we opening up the door for all sorts of privacy violations?
    3. I don’t trust anecdotal evidence that is posted to basically increase consumer traffic (by having patients posting their experience with their doctor online). These are filled with all sorts of bias, and can also be falsified in order to boost the esteem (and cash flow) of an undeserving doctor/provider. This, is as dangerous as commercials.
    4. Personal Health Records that are online are also not protected by major privacy laws, such as HIPAA, so there are no legal repercussions when someone signs up with one. I also question the ability of these PHR companies to adequately protect my privacy from theft and exploitation.
    5. A good search only requires knowing where to look. Every state has public universities. What is unknown to most people, is that all residents of their state (and some non-residents as well) legally have access to their entire state university library system, including their online library databases. All one needs to do is to take the time to sign up for access. These are some of the best resources the country has to offer.
    Overall, a good exploration of relevant issues, but I would say that your stated desires would require a more innocent, benign world. Kudos for the topic.