According to Star Trek, 23rd century doctors will diagnose most illnesses and determine the general state of health via a device called a Tricorder. But as with many of Star Trek’s futuristic technologies, aficionados of the show don’t want to wait for their great-great grandchildren to experience its wonders. Channeling Willy Wonka’s Veruca Salt, they chant, “I want it now.”
The XPRIZE offers one path toward pushing science fiction technology into the now—and the Qualcomm-sponsored Tricorder XPRIZE recently awarded $2.6 million to Final Frontier Medical Devices for their design, which was on display at the recent Silicon Valley Comic Con where Steve Wozniak received a private demonstration.
Basil Harris, an emergency room physician with a PhD in engineering, and co-inventor of the winning design, shared during a recent interview that Final Frontier Medical Devices was more a team name than that of a company. “The team name came from us asking ourselves who a starship supply officer would call for Tricorder replacements. We decided that the name wasn’t going to work for an ongoing concern.” Dr. Harris and his brother George, the company’s CTO, ended up founding Basil Leaf Technologies to further develop and lead the product through clinical trials.
The product the brothers produced ultimately became known as DxtER (pronounced “Dexter”), an artificial intelligence-based engine that diagnoses medical conditions by integrating clinical emergency medicine experience with data analysis from actual patients.
XPRIZE’s website sums up the goal of this competition very succinctly: “Imagine a portable, wireless device in the palm of your hand that monitors and diagnoses your health conditions.” Full guidelines can be found here. Here’s an early glimpse at DxtER along with some insight on its development so far.
The brothers worked together on this project, starting with a kind of deconstruction of Basil Harris’s brain into a rudimentary expert system that could be used as a guide for hardware design.
He explained, “When we built the first iteration it was ‘if-thens,’ like a traditional AI. We sat down for a couple of months and decoded, as a starting point — picked about my brain and tried to decode that. We were very data-driven.
“From there, we validated that model by gathering hundreds of charts of patients (approved through the institutional review board) with actual conditions we were looking for, and matched them to patients of the same age—we ran a match-case control study to see how well our initial model would do. It did well in some things and poor in others.”
The software evolved through several iterations of testing so that it can now make fairly reliable recommendations about conditions and next level tests.
DxtER: The Hardware
DxtER’s core hardware consists of an iPad mini, but it will run on any current smartphone or tablet. In order to retain the non-invasive nature of the fictional Tricorder, the team invented the DxtER Orb, a digital stethoscope for the analysis of abnormal lung sounds that combines with an infrared temperature sensor.
Other DxtER sensors include:
- Wrist Sensor for continuous monitoring of pulse, oxygen, heart rate and blood pressure, along with non-invasive monitoring of glucose, hemoglobin and white blood counts.
- A Chest Sensor for continuous monitoring of electrocardiogram (ECG), heart rate and rhythm, and body temperature.
- A Spirometer performs pulmonary function tests.
- A conventional Blood Pressure unit provides calibration.
The tablet guides the user through all the steps needed to collect data with the sensors. Medical staff obtain additional objective information through unique routines on the tablet, such as a vision test or neurological evaluation. The app further guides its users through step-by-step procedures for preparing urine and blood samples for independent analysis conducted outside of the system if necessary.
Batteries power the individual components, each of which is connected to the tablet via Bluetooth. The Tricorder stores analysis and results locally. The system remains operational without an Internet connection. If connected to the Internet, the Tricorder also sends results securely to an Amazon server.
The team built everything from scratch for the prototypes (they built 65 complete kits for testing at UC San Diego). Harris shares that “as we move forward we don’t plan on building everything ourselves. Instead, we will focus on developing the unique components further.”
This version isn’t going to determine species by running noninvasive DNA tests, or performing multivariate analysis to determine differences between a group, but the sensors, along with examination data, provide enough information to accurately diagnose several conditions, including:
- Urinary Tract Infection (UTI)
- Atrial Fibrillation (AFib)
- Sleep Apnea
- Chronic Obstructive Pulmonary Disease (COPD)
- Ottits Media
- Absence of Conditions
The system also offers preliminary indications for stroke, tuberculosis, hepatitis A, Pyelonephritis, Bell’s Palsy, Leukopenia, Asthma, Bronchitis, Dehydration, Ketosis, Hyperglycemia and Hypoglycemia.
In order to keep DxtER as non-invasive as possible, Harris said, “we use a technique for determining glucose, hemoglobin and white blood count without drawing any blood at all. We were pushing for that because it was like a true Tricorder…it’s not as accurate as sending a sample to a full chemistry lab, but it gives a sense of the range you are in so you can make decisions about which follow-up tests are necessary. We are taking that sensor into clinical trials now. We’ve advanced it a bit from the version used in the XPRIZE competition.”
From Theory to Practice
Dr. Harris stated that he doesn’t “think the first Tricorder kits are going to be fully automated as they were in the competition. I think they will provide an interface and some guidance to the medical provider.”
Clinical trials are crucial for the Tricorder to become a functional part of a medical tool kit. Because of the non-invasive design, FDA approval requires less rigorous testing than something put into the body.
Harris added, “We are using it in a hospital setting where people consent to be part of the trial. We are already drawing blood, for instance, and we use that to compare the results from the sensor.”
We may not need to wait for the 23rd century for a Tricorder, but it isn’t going to be a common instrument anytime soon. Like many scientific advancements in an ethical society, delays come not because of basic technology limitations, but because of the inferences and recommendations made by the software that need to be thoroughly tested. “Testing a single sensor is much easier than getting the entire Tricorder [approved] because it isn’t, as the Tricorder does, making a diagnosis and calling a person to action to do something about a condition. How do you regulate something like that? The FDA is trying to figure that out.”
As Dr. Harris says, “It has to earn our trust before we let it loose.”
Harris and his team, and others inspired by science fiction properties like Star Trek, aren’t waiting for the future, but as Alan Kay famously suggested, inventing it. Clinical trials continue at Lankenau Medical Center of Main Line Health.
A second-place prize of $1 million was granted to Taiwan-based finalist, Dynamical Biomarkers Group, led by Harvard Medical School Associate Professor Chung-Kang Peng, Ph.D. and supported by HTC Research.