One of the reasons the science has always been so interesting to me is that even after years of studying its various branches, in more or less detail, it never lacked the "wow" factor. Think CERN, the newly discovered "faster than light" particles, think miles and miles of underground equipment whose price tag would make GDP of some countries look sad. In my field, think state of the art detectors and algorithms that allow for live imaging of humans during surgery and allowing surgeons to see the images of tumors they need to take out from a patient and distinguish safely between healthy and diseased tissue. However, every now and then I am reminded that you can do amazing science even without all this expensive equipment; that is to say, the human ingenuity does not have a price tag. 

I sat in on a neat talk given to my department a week ago, and I thought it'd be a good topic to share to this audience. The talk was given by a professor from the College of William and Mary, and it was named: "Electronic Detection and Diagnosis of Health and Illness of Premature Infants". I'll try and give you the gist of it in a paragraph or two and perhaps some of you will find it as amusing as I did when I heard it. As I notice from some domestic news reports, it seems like the topic of improved care for the premature infants is somewhat "hot" with the ongoing "Sve za bebe" campaign.

The basic idea and drive behind this ongoing work is to see if one can help the ICU staff and MDs predict onset of sepsis for the prematurely born infants in a non-invasive manner. Dr. Delos and Dr. Moorman (MD) from UVa argued that there might be some patterns in the ECG that could point to higher risk of sepsis in infants, and started studying those patterns more carefully. Keep in mind that babies have healthy hearts in general, so the ECG is not used for checking for heart diseases - but rather for checking its periodicity and patterns that might be precursors for sepsis. The talk started off with a very intriguing visual brain teaser that confused the non-medical audience and can be found on Dr. Delos' webpage here. In this image, there are four samples of ECG heart rate records of infants, and the question posed was to judge the health of each of those babies. Surprisingly, the healthy baby was the top ECG, which looks the least periodic! It turns out that periodicity is not a good thing, as the baby should have a somewhat irregular heartbeat over long periods of time, that is the result of interacting with the surrounding, having nurses over to check their state, hearing noises and getting excited over them or napping for a bit.

Looking at the already existing ECG data from babies who were diagnosed with sepsis and connecting it with the timing of the diagnosis in those infants, the team concluded that two patterns imply higher risk of sepsis - reduced variability in the ECG as well as the increase in number of so called decelerations - transient slowing of the heart followed by a return to normal rhythm. They used these two traits to develop a figure of merit that would provide doctors with info as to which babies have a higher risk of sepsis at that moment. This figure of merit starts rising statistically significant almost a day before the sepsis event, which provides sort of an alarm as to which babies should be watched over carefully in the near future so they can be given the appropriate treatment before it's too late.

The group has provided their software (it can be loaded to the computer that is already used for observing and recording the ECG) for testing in eight hospitals across the US, and have seen positive and statistically significant trends in mortality rates so far. The post-talk questions revolved around interpreting data and discussing the setup of the study, which is pretty common for this type of seminars. The nay-believers quoted that any new equipment usually gets better results than it should due to being favored by doctors in a new clinical study (the new toy effect). In my opinion though, the fact that these investigators used a feature of system that is already in place, with very little added cost and complications to the user, is what makes me even more excited about research and learning as I have been as a teenager.