So the weekend is finally over and personally I wish it never did. That is how much I enjoyed the Random Hacks of Kindness Hackathon and being around every individual that participated in the event. Over the course of two days, I had the privilege of working with a team of remarkable individuals to help me bring my idea of a wound classification system to life. Although the weekend ended without finishing a working prototype, our team won 3rd prize based on our conceptual design.

First prize was awarded for Messages without Connectivity communication model, that allows the exchange of messages when cell towers are down (absolutely remarkable). I remember participating in recent discussions with a few of my friends in Syria that wanted to figure out how to do this in case the government cut down communication to isolate the protesters, and it is great to see that a group of my fellow hackers were able to create a working model. The second prize went to Bacon, an alert phone app that sends alerts once activated with the individual’s whereabouts.

So, to talk specifically about my team’s project, we were working on developing a system that allows volunteers to take images of wounds in a refugee camp, add demographic information to that image, then send the info to a central server where physicians can view the images, triage the wounds, and send back instructions to volunteers. Below is the flowchart we developed for this process:

Wound Classification Process

The reason why we weren’t able to have a working prototype by the end of the weekend was due to the fact that we bounced several ideas back and forth about the specific platform to use for such an app and went further to test those ideas from a technical point of view. Originally, we wanted to build our solution on top of an existing solution (FrontLineSMS), and use MMS to send the images and patient demographic information, but we later dismissed this idea for two reasons: 1) Not all carriers around the world provide MMS services, and 2) If we send the image via MMS, we had no way of indicating the coordinates of the wound edges, which we use in calculating the dimensions of the wound, and based on that calculate the number of stitches and estimated time to heal. This would have forced us to figure out an algorithm to automatically detect the wound in the image.

Next, we worked on using a combination of a phone app and a web-based interface for the physicians. The objective of the phone app was to enable volunteers to indicate the edges of a wound, and have the phone app record the pixel coordinates of those edges, which will be used in calculating the size of the wound. We had two programmers working simultaneously on developing an Andriod app and a BlackBerry app, while the rest of the team were working on a PHP interface. This was proposed to allows us to implement an API that does not limit us to one software package (as planned originally) but allow multiple systems to connect to our solution.

As a part of my pitch on day 2, I talked about a few future considerations for the app that I would like to share with you:

  1. Expanding the triage concept beyond wounds: As a part of a conversation I had with an emergency expert that attended the event, I had the opportunity to learn that wound triage is merely a small part of disaster triage, and that there are far more things to look for (e.g. internal bleeding, broken bones, etc), and those can be specific to a body part, and vary based on age. Perhaps a future expansion opportunity for the app would be to include that specific info in the form of short surveys that are easy to fill out by the volunteers, and send to physicians located remotely in order to help them with triage.
  2. Use of Teleconferencing: If we implement the idea above, we might not always get all the info that a doctor needs, so using a camera phone that supports teleconferencing can allow the physician to give volunteers instructions in real time to perform further triage and collect more information that help in making a decision.
  3. Automated Identification of Wounds: This was an idea that we talked about initially, by using advanced image processing capabilities, we could have the software automatically identify the wound in an image, but the algorithm was too complex to develop over the course of the weekend. However, in the future, if we can accomplish this, we will reduce the number of clicks, as with the current solution, the volunteer, after taking the picture, must specify the top, bottom, left, and right points of a wound, based on which, we calculate the dimensions.
  4. Automated Classification of Wounds: Using a reference image library of properly classified wounds, we can develop an algorithm that compares the image we take of a patient’s wound against those in the library in order to automate the classification process, and save time for physicians.
  5. More Accurate Calculations of Stitches & Supplies: Capitalizing on the idea of automating the calculation of stitches required, perhaps once we have enough information, we can write an algorithm that will take into consideration the dimensions of the wound, how long it will take to heal, and calculate all the other supplies required for caring for such a wound.

Like this post? Subscribe to my RSS feed and get loads more!