Epione

Using the latest in protein detection methods and image processing algorithms, this research could lead to a first-ever clinical tool to diagnose prescription opioid addiction.

One of our close family friends was injured in a car accident. As part of recovery, she was prescribed opioids as a painkiller. Our friend is one of 1,000 addicts who are treated every single day in the US, for the misuse of opioids. In the past year, the opioid addiction crisis accounted for 70,000 lives per year, with an economic cost of over 500 billion dollars. Most patients become addicts without even being aware of it. It’s an urgent and current social, economic, and medical problem that needs a solution as soon as possible. 

The project originally started with me trying to understand the opioid addiction health crisis and the possible role of science in dealing with it. I had a separate goal to study the role of certain genes in humans and protein detection techniques. Currently there are no tests to detect addiction. Normal patients who are given painkillers for reducing pain after surgery or other ailments can descend into addiction without being aware of it. This has resulted in a public health crisis and is currently costing thousands of lives each year. There was a need for a diagnostic solution, even if it is directional. 

With the advancements in protein detection mechanisms and colorimetry-based testing, as well as the involvement of gene sequencing in learning more about the mu-opioid receptor gene,  I believe that I can make a small difference in the fight to address the problem. The concept of early diagnosis is one that often gets lost in a plethora of treatment and prevention tools, but once research is complete, this tool should be the first ever tool to clinically diagnose for addiction with a blood or bodily fluid sample. In contrast to existing methods such as surveys and psychological assessments, my device aims to be a medically accurate device to easily perform a screening before a patient is prescribed a medication or to measure progress in a rehabilitation center due to the correlation in protein production. As I progressed in my work, I have further narrowed it to early detection, which I am sure has the potential to save many lives, including our friend. 

I combined both my goals when I was presented an opportunity to shadow in Dr.McMurray's lab. As the project progressed I sensed an opportunity to also develop an inexpensive and portable solution that replaces traditional expensive spectrophotometer and can provide accurate directional results to both physicians and end-users together. My hypothesis was that if I can show a correlation between protein expressed by the “addiction” gene OPRM1 and the amount of opioid in the system, I can probably build a calibrated scale which can be shown on a user interface.

Prescription opioid addiction is a national health crisis. Today, patients or physicians don't have a way of diagnosing if a painkiller given for relief causes addiction. Current approaches for diagnosis are based on self-assessment or psychological evaluations.

The proposed solution measures variations in the Mu Opioid Receptor protein produced by the OPRM1 human gene. The experimentation included the behavior of OPRM1 gene in response to exogenous opioids such as prescription drugs (oxycodone/ fentanyl). In addition, the increase in protein levels due to agonists was mapped to a user-friendly scale for physicians to take action. The research output also included design, fabrication, and testing of a portable-prototype tool to indicate the onset of opioid addiction in patients directionally.

My work involved simulating the behavior of human genes addicted to opioids using the human OPRM1 gene on a Saccharomyces Cerevisiae host. The methods used included yeast-strain preparation with CRISPR/cas9 expressed with OPRM1. The portable solution was calibrated to substitute the colorimetry process of protein detection with neural network-based image processing algorithms. The results are sent over Bluetooth to a custom-developed mobile app, where they are mapped to a user-friendly scale and displayed for further action by the user.

Since the solution is still at a prototype stage, there is no current population that I’m working with. However, when the product is released, the main customers would be physicians, rehabilitation centers, and primary health care centers with a future goal of being able to perform at-home tests that patients can use, 

Currently, there are no devices or tools for doctors to monitor levels of addiction before prescribing opioids. Additionally, there are no tools for rehab centers to keep up with the status of the patients.

To tackle this problem, Epione is an Inexpensive and portable solution that replaces expensive colorimetric testing mechanisms. Instead, it can provide easy monitoring techniques for physicians for scanning before prescription and rehabilitation centers for keeping track of progress.

As an avid innovator that creates through the ideology of synthesis by putting things together rather than analyzing, breaking them down, and working on individual components, my work depends on failure and bouncing back from mistakes. I’ve built many devices and prototypes and my processes, with the biggest and most important step being “brainstorming,” where I focus on quantity over quality. This step stops me from stigmatizing failure and allows for a plethora of ideas that can then narrow down. Due to my innovation experience and willingness to fail and learn from my mistakes, I demonstrate my commitment and dedication to a specific project. A good scientist and “maker” requires a sense of determination to stay with problems longer and create solutions utilizing an iterative development of ideas. 

However, an innovative skillset and positive mindset must work with others. A concept that we consider design thinking is increasingly giving way to solving bigger problems rather than focusing on “organic designs” or “colorful user interfaces.” Recognizing this shift in the innovation movement, I decided to use design thinking to narrow down bigger solutions and combine different fields to allow for novel technology and ideas. This multi-disciplinary approach towards ideation builds more creative solutions, putting things together as nobody has ever seen before. 

By persevering through challenges, normalizing failure, and connecting different fields to maximize innovation and creativity, I carry the mindset of an avid innovator and scientist.    

Since the age of 7, I’ve built a variety of different solutions to tackle global problems. I now have a patented device to detect for lead in drinking water using carbon nanotube sensor technology. Working with the UNICEF Office of Innovation, I’ve built an anti-cyberbullying service using Natural Language Understand and Processing to detect words and phrases that may be considered cyberbullying on a variety of different front ends. I prototyped a genetically engineered bioelectric transducer to detect for Cryptosporidium transposons in drinking water. This past summer, I worked with the Koch Institute to create stronger lung tumor drug delivery mechanisms and I worked with the Broad Institute to optimizing gene sequencing technology. These ideas along with the partnerships and connections I’ve made with the industry demonstrate not only my scientific passion and creativity, but my initiative and willingness to bring ideas to reality. 

While it was exciting to learn new tools, technologies, processes and subject areas, the biggest thrill for me was to find out that such a solution was possible and can be validated through test results.

Going through the research process, I had to change the way in which I approached specific solutions. I like to work across technologies and use the “solutions looking for problems” approach by recognizing technology that may have the potential to tackle specific problems that I find a connection to. Especially since my work involved multiple fields and was a cross-disciplinary solution, I had to seek guidance and papers in cell biology, genomics, machine learning, mobile development, 3D design, and even the legal implications of opioid addiction. My interest in STEM was further increased by the recognition of what I should expect and how I should approach the research process especially as a teenger. I was able to engage in high quality labs, the support of menotrs, and validated research material as well. Along with my particular project, my interest has recently expanded to the applications of genetic engineering in energy, environment, and sustainable living- a big area of growth for our country.

A good scientist and engineer cannot just hold an idea without any way of communicating it to the customers and target market. After developing early versions of my prototypes and validating results with the lab that I worked at, I immediately moved to the next step of communicating with the target market and the Colorado Department of Law under the opioid response program. As part of this, and further communications with the Schreiber Group, a legal action group on the opioid crisis, I was able to learn more about the lack of opioid addiction diagnosis tools and the gap that I would be filling if my clinical testing device was available for physicians to use. Once determined, I went beyond just a scientific experiment and formulated a business plan and revenue model that I shared with investors and organizations to seek for support, feedback, and funding to continue research.

The initial prototype created was shared with several Denver area physicians and Dr.Eric Strain, John Hopkin University director of substance abuse, to get initial feedback. The feedback provided showed that there is a need for a reliable device for physicians and the device like Epione can help with that. The feedback also showed that if there was an easier way to use an enzyme-free testing, it would be a quick initial test for the physicians and then the samples can go for further clinical testing. 

During the process, I also reached out to the Schreiber group, whose mission is to influence public health policy. The founder of the group Terri Schreiber herself went through opioid dependence, and she supported such a device that physicians can depend on.

I reached out to the closest addiction center or rehabilitation center to understand the diagnosis tools they use, and most of them were self-assessments or behavioral assessments and lacked any major clinical way to prove before they started therapy.

Current research recognizes the behavior of the mu opioid receptor or OPRM1 gene in response to exogenous opioids. Leveraging this research, I attempted to find a correlation between the concentration of protein produced by the mu opioid receptor and a status of addiction to prescription opioids. Identifying this correlation allowed for the easy development of a platform that can provide directional results as to a status of addiction that physicians can use to diagnose patients. 

However, in addition to the current work, the concept of a correlation as established by the research for Epione can be utilized for detection of alcohol intoxication and addiction to recreational drugs using other gene markers in the body. For example the tao opioid receptor, which shows indications of alcohol addiction may also demonstrate a similar correlation with the protein that it produces. 

In this manner, not only are we able to diagnose for prescription opioid addiction at an early stage, but this idea can be further expanded to diagnose addictions to other substances. 

Apart from the protein correlation, another area of research that is innovative is the development of a portable spectrophotometer built with a Raspberry Pi microcontroller and 8 MP camera in order to cut images down to the region of interest, convert them to grayscale, and then additionally send the results to a phone over Bluetooth. 

Current spectrophotometers in labs to measure optical density of samples, especially those are products of ELISA tests cost approximately $15,000 and instead can be recreated using a portable microcontroller as utilized in Epione.

My impact goal for the next year is as follows:

Today no easy diagnostic tests exist and diagnosis relies on the patient's self initiative. This is too late for many patients. My goal is to work with the Colorado Dept of Law to create awareness of early diagnosis of prescription opioid addiction. It will be some time before my solution will be available, but we can save lives by encouraging patients who are on prescription opioids to voluntarily check for diagnosis of addiction. 

In this next year, I hope to pioneer the genetic engineering approach of diagnosis, continue research and enhancements of the app, and gain guidance from professors and researchers of elite universities.

Beyond this year, I have significant plans to grow my solution further. I have four key objectives that I would like to meet:

• Commercialize the solution

• Gain a +30% Revenue Year over year

• International Expansion [20% of customers outside of the United States]

• Producing the #1 Biomedical app 

ELISA, also known as Enzyme Linked Immunosorbent Assay, is a method of protein detection which utilizes a colorimetry-based approach to identify concentrations of proteins specific to the antibody that is tailored towards them. In this particular solution, the OPRM1 antibody is used to detect concentrations of the mu opioid receptor protein, the HRP enzyme is added to alter the color depending on the concentration of the protein, and the results and optical density of the color is then read through a spectrophotometer for quantitative results that can then be mapped to a specific “addiction status.” When performing ELISAs in the lab, they prove to be the most sensitive method to pick up purified protein concentrations with the most user-friendly results that can be interpreted through various technological tools. 

One of these examples, which is also a core component of Epione, is the neural network core of the device. During the original training of the device, over 50 different sample images were added into the Raspberry Pi which was running a convolutional neural network algorithm, leveraged from previous research conducted by a University in India. These images were then cut down to the region of interest and processed to create baseline images for the network. These results were then tested and calibrated to create a full set of images that are each individually correlated to a status of addiction using a strong machine learning algorithm. When a sample is tested in the device the high-resolution image is fed to the neural network algorithm to predict a new value based on trained data. This result is mapped to a scale on a companion mobile application developed in MIT AppInventor,  which then provides directional results to a physician on a 5-color scale.

Currently, this solution is still at the prototype phase, so it doesn’t serve any patients. However, through science fairs, the solution has been introduced to 2500 students and researchers, and through talk shows and speaking opportunities, the solution has been spoken about to approximately 20,000 people. 

Within the first year of launching, the goal is to directly serve 1,000 consumers with 50% of the market being doctors and primary health care centers, 30% being addiction relief and rehab facilities, and 20% being directly to the patients, once research is complete on self-testing.

Following are some of the barriers of the research at the time of my work, most of which can be addressed with the right resources. 

1. The yeast strain was specifically engineered to replace the STE2 gene with the OPRM1 gene to perform a protein analysis due to ethical constraints. However, all testing was performed on a Saccharomyces yeast strain which doesn’t behave exactly like a human with the mu opioid receptor gene. To accurately bridge the gap between yeast and human test, testing on mice models must occur 

2. To build the artificial neural network, a few test images were added into the program. However, to build a completely accurate neural network, hundreds of thousands of sample images must be added. Due to time constraints and the lack of sample images, only about 100 data points were added to build the neural network, not creating the most accurate one but allowing for space for improvement 

3. The OPRM1 antibody was commercially purchased and not individually manufactured due to the extended amount of time that would take. Since it was purchased from an external source, the validity and accuracy of the antibody cannot be verified, potentially skewing the results and not attracting as much protein as possible. 

4. For the project, I incurred the costs from my own savings. For scale testing there is a need for better financial support due to significant costs of reagents and agonists. 

5. In order to continue testing, I need to be able to have continued access and approval to labs. Specifically BL2 and BL2+ laboratories, that allow for future phases of testing with mammalian models such as mice.

Colorado Department of Law Opioid Response Program (Attorney General’s Office) - As part of an internship program in the fall of my junior year, I partnered with Colorado Office of Attorney General's to support the state's opioid response plans. Through this I contributed to strengthening policies and laws that support community activism to ensure lasting impact of changes in our communities. Along with this I shadowed Dept of Law officials on campaign management and target demographics to create a successful opioid response plan for the salient challenges in our community. By learning more about the legal actions that can be taken for opioid response and the target demographics, it allowed me to learn more about the problem at hand. 

Terry Schreiber and The Schreiber Research Group - By contacting the Schreiber Research Group, a group that works specifically with research and fact-based evidence regarding opioid addictions, I hoped to find gaps in the current solution that existed. Through this interaction, I gained that there were treatment solutions at rehabilitation centers and prevention techniques through educational courses but a big knowledge gap in the way that prescription opioid addictions were diagnosed in a physician setting, the area which I decided to focus my work in.

My business model looks to serve millions of patients around the country who are prescribed opioids for pain relief. I estimate the cost of a test today to be $150 per unit with access to the free app. 

Within the first year, I hope to reach 1000 target customers. In the second year, I hope to reach 5000 customers, and 10000 in Year 3. Through this, in year 1 I hope to gain $150,000 with a $150 unit cost, $675,000 (4500 units) in year 2, and $450,000 (3000 units) in year 3. 

I have a few key goals that I want to meet to provide the best version of the product to the consumers: 

Research - 

• Calibration: standard colorimetry device and the application 

• Confirmation: HIPPA and Security Administration

• Feedback & User interviews: Rehabilitation Centers and Physicians  

Team - 

• Medical device UX designers 

• Collaboration with John Hopkins, University of Colorado and Harvard professors for feedback on my research

• Partnerships with community organizations, rehab centers, and physicians

• Working with FDA

Manufacturing - 

• Partnering with electronic device manufacturers 

• Scale testing with colorimetry devices

• App confirmation and approval in the google store

This financial plan demonstrates how money is gained over three years through bought capital and income. Through an investor for three years, we gain $185,000 every year with $25,000 being self-funded. Through the sale of individual units, we hope to gain $450,000 by year 3 and $700,000 through licensing by year 3. Considering the operating costs, the profit will be approximately $680,000 by year 3. 

When recognizing growth opportunities and ways to target customers to gain revenue, there’s a few key elements our company would like to focus on: International Expansion, Partner Expansion, and expansion of the tool to other adjacent tests that require spectrophotometer usage. Along with this, we plan to slowly release to all 50 states in the United States after conducting preliminary clinical trials in Colorado and moving to the international space. 

Secondly, when going to market, the elements that require focus are: Working with FDA, App store, Licenses, Paid search, Industry Trade Conferences, and Awareness campaigns.