Monday, March 23, 2009

Now Talk to Twitter Through Personalized and Responsive Information Interfaces

The Twitter based APP released by TrialX delves into a potentially huge but relatively unexplored aspect of Twitter as a medium of communication. That is, the potential of Twitter to be a medium to enable users to not just "send updates" but a way to "push" tailored information to them based on their tweets.

The TrialX/Twitter APP demonstrates that one can "Talk to Twitter" or in more general terms one can visualize it as a potentially powerful 'Q and A' medium. From a data and information availability perspective, the TrialX/Twitter APP demonstrates that one can use Twitter as an interface to information locked in databases of thousands of sites or (excitingly), Twitter could be extended to become a platform for 'Personalized and Responsive' Information Agents(PRIA).

Lets see how? The TrialX APP allows you to find out which clinical trials are running in your state personalized to your health condition (e.g., diabetes), gender and age by simply sending a QuTweet (a query tweet, pronounced as cute-tweet) to TrialX. So if you send this QuTweet to TrialX , '@trialx CT i am looking for diabetes trials for 45 yr old male" ,you will get an instant twitter reply/response providing you a link to the matching trial results on TrialX (see more examples and usage details).

Some interesting and innovative aspects of the APP are

1. It inverts the commonly used communication paradigm on Twitter;instead of updating, one is now asking and "getting" information back

2. The Information provided is personalized based on user input. This is different to say, subscribing to a Twitter user (channel) for say latest information about clinical trials or a general health information channel. In the latter case one receives updates that are generic. The TrialX APP only sends personalized responses

3. The information is ON-DEMAND (with a minutes delay on average) and makes it appear that Twitter is "responding" to the users information need

4. The information provided is automatically generated and not a response from another human user. This makes it very interesting as one can envision many such automated information agents

5. Importantly, the information comes from a third-party database and not from the Twitterverse. This makes it at-least theoretically possible to envision twitter being a search Interface to millions of third-party databases (would this be a better way to make the deep web searchable, who knows?)

Apps based on the above characteristics of the TrialX APP can be envisioned (sure there may be already some existing. I found one APP for finding book prices based on tweeting the ISBN number). Here are some examples listed as illustration of the idea:

1. Send a QuTweet to a physician directory service to find a local specialist, 'Looking for a top-rated orthopedician in zip 10004'. Or send a QuTweet to a Hospitals Twitter Channel to find the hospital's specialist, 'Tell me your cardiologists'.

2. Send a QuTweet to a food site's twitter channel to find recipes. For example, QuTweet this to '@ifoodtv chicken recipe, indian style' could return you a link to a page containing chicken recipes, indian style, from ifood.tv.

3. Search flight tickets by QuTweeting to say Kayak, thus, 'need ticket for round-trip from new york to LA between 03/23/09-03/30/09'. This would return a link to the search results on kayak

4. Obtain matching profiles from online dating sites like Match.com by QuTweeting, "Looking for females between 27-35yr old in NY"

Twitter's updates being less than 140 characters, actually works as a great advantage here because parsing these natural queries for a specific domain/site is not that hard. With some patterns (and good user examples), its easy to pick up most of the queries.

Such APPs could work by responding to users in two modes. One, by instantly responding to individual QuTweets sent by users, such as the TrialX/Twitter APP. Two, they could push personalized information resources on a daily basis, based on scanning all the tweets of a user during the day.

The above APPs could also become a monetizing mechanism for Twitter. Since each QuTweet provides a link to the third party's site with the search results, Twitter could charge a "referral fee" for each transaction or charge on a cost per 1000 referrals. Since the users intention is clear, these referrals would likely be potential customers. For a third-party website, this could potentially be better than Google AdWords for several reasons (will list them later) other than the fact that this interface provides better search results based on a dynamic database query and not based on pages being indexed and linked back from Google

Guide to Building Healthcare Apps for Twitter: Some Categories of Healthcare Apps

I recently listed some some characteristics of communication enabled by Twitter. Based on these, there are several different categories of uses that can be envisioned for Twitter in relation to healthcare.

1. Information dissemination and Promotion. One of the obvious and most commonly uses of Twitter is as a medium for promotion and marketing. Hospitals, doctors, or health and wellness programs can disseminate information quickly through simple tweets. These could be tweets that provide general health information, or tweets about activities at an organization or even information specific to an individual or organizations. Several examples exist and include the MD Anderson http://www.blogger.com/img/blank.gifand Henry Ford Hospital which are engaging consumers through their twitter channels (@MDAndersonNews and @HenryFordNews respectively). The Henry Ford Hospital recently broadcast a live operation through tweets (demonstrating the use of the "now" characteristic of communication enable by Twitter).

2. Consumer Health APPs
a. Personalized and Responsive Search Engines (PRSE)- these apps provide personalized information resources based on a query tweet. An example is the TrialX/Twitter App that allows a user to QuTweet, say ' Looking for a diabetes trial for a female, 45 in NY' and the app returns a response with a link to a personalized search resulhttp://www.blogger.com/img/blank.gifts page on Trialx listing the matching clinical trials. See more details of PRSE

b. Self-quantification and monitoring APPs. Twitter users usually tweet with updates of their daily lifestyle activities that impact health behaviors such as diet, physical activity or smoking. Examples like "eating a McChicken", "heading to the gym", "got off a yoga class". These tweets or "digital lifestyle crumbs" can be monitored to provide lifestyle behavior modification recommendations. For example, sending a tweet back saying "did you know that a McChicken has 300+ calories" or providing an alternative suggestion to replace what people eat. Similarly, people can be motivated to exercise more or to quit smoking or a personalized link can be sent to the user containing list of people on twitter who are talking about quitting smoking (see twitter characteristic about propagating information along social networks).

3. Research Dissemination. One of the key challenges facing scientific research is dissemination of research findings into routine clinical practice (commonly referred to as the T2 translation block). Or even beyond to the general population. Several characteristics of current knowledge dissemination methods underline the need for new communication modalities. One, there is a considerable time lag or latency (in years) between the conduct of scientific work and it’s reporting, both to the scientific community and to the general public. Second, current methods of dissemination are primarily based on academic publications or via materials such as software code or tools, which have limited audience. Third, published work is more descriptive of “what was done and achieved” but lacks details of work as it happens and the decisions that are made in the research process.

Twitter can address these shortcomings - it is by default meant for "communicating the now". And second it reduces the barriers to data entry, allowing quick updates, of say , daily research activities to be communicated as they happen.

These are some just categories of potential use and in the next post, I will discuss APPs in the area of In-patient care, medical education and others. And ofcourse, will look at some pitfalls of the medium too.

Listing of some uses of Twitter in Healthcare As mentioned elsewhere

1. Using Twitter for Health Care by G. Antwerp
2. 140 Uses By Phil Baumann

Guide to Building Healthcare Apps for Twitter: Consider Twitter Communication Characteristics

As Twitter diffuses further into the mainstream (with even senators twittering to their hearts content), it is beginning to find its way into Healthcare. There are several users twittering about personal health and wellness information. Just recently there was a story of a surgeon twittering a live operation.

I have being using twitter since 07, mostly for personal reasons but have recently started exploring twitter as a medium to disseminate information about two healthcare related projects. This experience has provided insights about some key characteristics of communication enabled by Twitter which makes it an attractive tool for healthcare (Note: these are purely based on observations and not validated through scientific studies. And by know means are is this complete). It may be worthwhile to consider the following poitns while building apps for Healthcare.

1. Communication in twitter is time-sensitive (‘what’s happening now’), making it inherently suitable for communicating things as they happen.

2. It reduces the effort needed to put information out since tweets are short

3. It makes communication “conversation-like” and also social since a users updates are broadcast to all their "followers". The latter aspect is particularly important in propagating health information along social networks. As per social-cognitive theory, people learn by observing others, and hence a positive health promotion message has greater chance of uptake when there is an endorsement from someone in your peer network.

4. The network effect of communication also has a potential for message “amplification”; a hub in the information chain can disseminate information to thousands of their “followers” instantly.

5. Very importantly, twitter updates can be seen as “digital trails” of people’s daily activities and thoughts. It is common to see people tweet about things like "heading to the gym" or "i just had a burger". These trails can be "scanned" to provide tailored information resources.

6. Twitter makes it easy to access the communication medium through several different interfaces on the web and hand held devices.

7. Twitter provides the ability to keep updates entirely private or if needed enable one-on-one communication through a direct message channel. This is important from a healthcare perspective because privacy concerns are real and legitimate.

Again, this is not an exhaustive list of communication characteristics enabled by Twitter but it can serve as a useful guide to envision uses of Twitter in Healthcare. Next, i will list some categories of such uses.