![Picture](/uploads/5/0/2/7/50279965/published/img-1921.jpg?1491749979)
This post is about my amazing weekend at the Texas Medical Center Biodesign Hackathon in Houston! I assumed "hackathon" was a commonly known term, but after a few conversations with non-techies (no mom, we're not formally gathering to commit unspeakable cyber security crimes) I realized not everyone is aware or has had the opportunity to compete in an awesome event like this. So quick lesson: "hackathon" is a derived from 2 words:
hack - hacking means clever programming that often modifies or alters standard computer software or hardware. We "hack" together a solution by using elements of this, components of that, and crafty use of that.
marathon - although the length of hackathons varies from a day to a week, the standard is focused, intense coding or hacking.
So that's the basic framework for a hackathon. It's honestly more of a sprint, but I suspect word-smithing of hack and sprint variations just hasn't yielded any portmanteaus that can compete with hackathon. The Hackathon Wikipedia also accurately points out that during such a session, " eating is often informal, with participants often subsisting on food like pizza and energy drinks" . I chuckle at the language- but it's true. While TMC provided us with awesome food, our team often forgot to eat, substituting balanced meals with pickles, Monsters, and K-cup fuel. Sleep is also informal - we worked until midnight Friday night, 4am Saturday night, and 5 pm Sunday!
This isn't some people's idea of fun, but for people passionate about creation, networking, and learning, a hackathon is an amazing experience. Moreover, this TMC hackathon was focused around healthcare! On Friday, we were presented with four unmet needs in healthcare - two in the medical device setting and two in digital health. After presentation of the unmet needs, we selected our need, formed teams, and began hacking! Meet my impressive team: Erik is CEO and Cofounder of the healthcare consulting company Calamine and came bearing rich experience in health information systems and user interface (UI) design. The amazing Shadi is a mechanical engineer and a research intern at the Texas Heart Institute, and augmented reality (AR), heat transfer, and fluid dynamics are just a few of her specialties. The talented Hessam is a PhD candidate in computer science and comes with an impressive range of skills including IoT (Internet of things) and distributed systems. Like I said, POWERHOUSE team.
The reason this weekend was so inspiring to me isn't just because I got to meet talented people like my teammates, it was the inspired creation! What we all shared was a passion for innovating in healthcare. Working 30+ hours over a weekend, you've GOT to have passion! (Or craziness. Maybe we're all just a bit unhinged. ;)
While my team didn't win the pitch competition, we were recognized for best tech. What we created was a real-time tracking system and analytics engine to optimize patient transport and scheduling to radiology. Getting inpatients to radiology is inefficient, in part due to all of the pieces needed including a prepped patient, equipment (wheelchair or bed), and transport staff. Moreover, there is no science to the scheduling of patients on the machines (X-ray, MRI, ulrasound, and CT). This is how we broke down the problem and solved it:
1) Problem: where is all of the stuff?
Solution: A real-time tracking system that locates patients, equipment, and staff through a ultrawide band (UWB) sensor system that can locate these assets within 5 cm. Because this system uses wireless sensors, it is low cost, high accuracy, and automatic - no manual entry required.
2) Problem: what about all of the areas where hospitals lose WiFi signal?
Solution: Image recognition technology reads the unique bar-code of each of the sensors. This technology means that even when signal is lost, sensors can "see" wheelchairs, patients - anything with this unique bar-code because the software has been trained to associate bar-code features with these objects.
3) Problem: Which route should the tech transport staff take in this sprawling hospital, where radiology is 7 flights down and in another wing?
Solution: Image recognition works in concert with augmented reality to point out the optimal route. As you'll see in the video, Shadi's phone recognizes directions to the ER and actually casts an arrow on her view that directs her to the ER!
These aren't just theoretical solutions - have a look at the 1 minute demo Shadi and Hessam created that shows how the technology works. Hessam walks around, and then we see EXACTLY where he went in this large warehouse with these long range, high accuracy wireless sensors. Next, Shadi has her phone and is walking around in the hospital, and the AR interface tells her which way to go and how long it will take.
hack - hacking means clever programming that often modifies or alters standard computer software or hardware. We "hack" together a solution by using elements of this, components of that, and crafty use of that.
marathon - although the length of hackathons varies from a day to a week, the standard is focused, intense coding or hacking.
So that's the basic framework for a hackathon. It's honestly more of a sprint, but I suspect word-smithing of hack and sprint variations just hasn't yielded any portmanteaus that can compete with hackathon. The Hackathon Wikipedia also accurately points out that during such a session, " eating is often informal, with participants often subsisting on food like pizza and energy drinks" . I chuckle at the language- but it's true. While TMC provided us with awesome food, our team often forgot to eat, substituting balanced meals with pickles, Monsters, and K-cup fuel. Sleep is also informal - we worked until midnight Friday night, 4am Saturday night, and 5 pm Sunday!
This isn't some people's idea of fun, but for people passionate about creation, networking, and learning, a hackathon is an amazing experience. Moreover, this TMC hackathon was focused around healthcare! On Friday, we were presented with four unmet needs in healthcare - two in the medical device setting and two in digital health. After presentation of the unmet needs, we selected our need, formed teams, and began hacking! Meet my impressive team: Erik is CEO and Cofounder of the healthcare consulting company Calamine and came bearing rich experience in health information systems and user interface (UI) design. The amazing Shadi is a mechanical engineer and a research intern at the Texas Heart Institute, and augmented reality (AR), heat transfer, and fluid dynamics are just a few of her specialties. The talented Hessam is a PhD candidate in computer science and comes with an impressive range of skills including IoT (Internet of things) and distributed systems. Like I said, POWERHOUSE team.
The reason this weekend was so inspiring to me isn't just because I got to meet talented people like my teammates, it was the inspired creation! What we all shared was a passion for innovating in healthcare. Working 30+ hours over a weekend, you've GOT to have passion! (Or craziness. Maybe we're all just a bit unhinged. ;)
While my team didn't win the pitch competition, we were recognized for best tech. What we created was a real-time tracking system and analytics engine to optimize patient transport and scheduling to radiology. Getting inpatients to radiology is inefficient, in part due to all of the pieces needed including a prepped patient, equipment (wheelchair or bed), and transport staff. Moreover, there is no science to the scheduling of patients on the machines (X-ray, MRI, ulrasound, and CT). This is how we broke down the problem and solved it:
1) Problem: where is all of the stuff?
Solution: A real-time tracking system that locates patients, equipment, and staff through a ultrawide band (UWB) sensor system that can locate these assets within 5 cm. Because this system uses wireless sensors, it is low cost, high accuracy, and automatic - no manual entry required.
2) Problem: what about all of the areas where hospitals lose WiFi signal?
Solution: Image recognition technology reads the unique bar-code of each of the sensors. This technology means that even when signal is lost, sensors can "see" wheelchairs, patients - anything with this unique bar-code because the software has been trained to associate bar-code features with these objects.
3) Problem: Which route should the tech transport staff take in this sprawling hospital, where radiology is 7 flights down and in another wing?
Solution: Image recognition works in concert with augmented reality to point out the optimal route. As you'll see in the video, Shadi's phone recognizes directions to the ER and actually casts an arrow on her view that directs her to the ER!
These aren't just theoretical solutions - have a look at the 1 minute demo Shadi and Hessam created that shows how the technology works. Hessam walks around, and then we see EXACTLY where he went in this large warehouse with these long range, high accuracy wireless sensors. Next, Shadi has her phone and is walking around in the hospital, and the AR interface tells her which way to go and how long it will take.
This was just a quick video for our pitch, but hopefully you can see how real, accurate, and EXCITING this tech is! There are companies out there trying to solve hospital operational efficiency, but our team's solution wins in terms of accuracy, image recognition, and advanced analytics.
What's really cool about this solution, is it would ALWAYS be collecting data. Nearly 24/7, we would be building a spatio (location) temporal (time) record of patient activity. As a data scientist, that feature is a goldmine because not only is the data self-generating, but the dataset is always growing, meaning insights will become more and more rich with time.
4) Problem: There is no science to radiology scheduling, with the patient order usually based on arrival time and urgency. Complexities such as transport delays, patient prep, and emergency cases that need imaging right away mean that expensive machines sit dormant while patient wait time drags on.
Solution: Use the spatiotemporal data from the wireless tracking system and electronic medical record (EMR) data to build an algorithm that optimizes patient medical image scheduling. This may sound complex, but it's similar to the global optimization that happens as flights are delayed or cancelled. When a cancellation or delay happens, the entire system adjusts to optimize airline time and money. In the hospital, the entire system is adjusted to minimize TOTAL patient wait time. There are thousands of ways to schedule, so it's not efficient to check every scenario. Instead, an efficient algorithm should check combinations until it finds one that meets certain criteria: ie emergency patients are seen first, 50 patients are imaged, and the imaging happens within 24 hours - for example. In my simulation, I planned to use an ant colony optimization (ACO) algorithm with a greedy randomized adaptive search (GRASP) to find optimized scheduling in real-time. See a little snippet of my python code here. I'm demonstrating that multiple scenarios are being tested, and when a scenario that fits the criteria is met, the loop stops. (Check out references below to learn more).
In close, that's what happens when four crazy, passionate creators get together for 48 hours and drink lots of coffee in an old Nabisco warehouse. I walked away from this weekend with a renewed sense of faith in the creative future of healthcare. Perhaps this system is reluctant to change, and there are more barriers and regulation than in other industries, but there are a slew of brilliant, resourceful entrepreneurs eager to tackle and solve complex problems in healthcare.
Sources:
1. TMC Biodesign
2. Using Tracking Tools to Improve Patient Flow in Hospitals
3. The best way to improve radiology patient transport efficiency
4. Study on GRASP-ACO Algorithm for Irregular Flight Rescheduling
5. Online Aircraft Scheduling with Ant Colony Optimization
6. An Irregular Flight Scheduling Model and Algorithm Under the Uncertainty Theory
7. Stanley - A RTLS Competitor (but we do it better :)
8. YouTube of our Demo!
9. Monte Carlo stimulation snippet I wrote in python
What's really cool about this solution, is it would ALWAYS be collecting data. Nearly 24/7, we would be building a spatio (location) temporal (time) record of patient activity. As a data scientist, that feature is a goldmine because not only is the data self-generating, but the dataset is always growing, meaning insights will become more and more rich with time.
4) Problem: There is no science to radiology scheduling, with the patient order usually based on arrival time and urgency. Complexities such as transport delays, patient prep, and emergency cases that need imaging right away mean that expensive machines sit dormant while patient wait time drags on.
Solution: Use the spatiotemporal data from the wireless tracking system and electronic medical record (EMR) data to build an algorithm that optimizes patient medical image scheduling. This may sound complex, but it's similar to the global optimization that happens as flights are delayed or cancelled. When a cancellation or delay happens, the entire system adjusts to optimize airline time and money. In the hospital, the entire system is adjusted to minimize TOTAL patient wait time. There are thousands of ways to schedule, so it's not efficient to check every scenario. Instead, an efficient algorithm should check combinations until it finds one that meets certain criteria: ie emergency patients are seen first, 50 patients are imaged, and the imaging happens within 24 hours - for example. In my simulation, I planned to use an ant colony optimization (ACO) algorithm with a greedy randomized adaptive search (GRASP) to find optimized scheduling in real-time. See a little snippet of my python code here. I'm demonstrating that multiple scenarios are being tested, and when a scenario that fits the criteria is met, the loop stops. (Check out references below to learn more).
In close, that's what happens when four crazy, passionate creators get together for 48 hours and drink lots of coffee in an old Nabisco warehouse. I walked away from this weekend with a renewed sense of faith in the creative future of healthcare. Perhaps this system is reluctant to change, and there are more barriers and regulation than in other industries, but there are a slew of brilliant, resourceful entrepreneurs eager to tackle and solve complex problems in healthcare.
Sources:
1. TMC Biodesign
2. Using Tracking Tools to Improve Patient Flow in Hospitals
3. The best way to improve radiology patient transport efficiency
4. Study on GRASP-ACO Algorithm for Irregular Flight Rescheduling
5. Online Aircraft Scheduling with Ant Colony Optimization
6. An Irregular Flight Scheduling Model and Algorithm Under the Uncertainty Theory
7. Stanley - A RTLS Competitor (but we do it better :)
8. YouTube of our Demo!
9. Monte Carlo stimulation snippet I wrote in python