Choose “NOW”
The Wondrous World of Medigene
By Valkyrie Holmes
What if I told you that in just a few hours, you could know all of the diseases you have and will have in your lifetime?
Come on, Valkyrie, that can’t be true. That stuff is only possible in science fiction novels and TV shows! Well, I’m here to tell you that it’s quickly becoming a reality and it’s happening super fast.
It’s all occurring through the power of gene sequencing and AI. This process of identification is a new and improved method of diagnosis and it keeps getting better and better every day. Not only have scientists figured out how to sequence all 3.2 billion base pairs in the human genome, but they’ve even figured out how to apply neural networks to create easier to read results for geneticists using artificial intelligence. Crazy, right?
Why should we care?
Sequencing can be used to monitor the development of both common and complex diseases like heart disease, diabetes, cancer, etc. It can also be used to spot markers for diseases that we don’t even know we have.
One of the biggest problems in the medical world today is that results for tests and treatments are super slow. Diseases like dementia and Alzheimers take months to diagnose. Many of the biggest killers in the medical industry could easily be stunted if they were identified early enough.
Let me draw your attention to the story of Sebastiana Manuel.
From the day she was born, she was haunted by terrible seizures, couldn’t eat, and was suffering from rapid convulsions. She was eventually rushed to the Rady Children’s Hospital in San Diego.
At first, the doctors had no idea what was wrong. They ran brain scans, MRIs, and transferred her to intensive care. After days of treatment with no results, the doctor told her parents that she wasn’t going to make it. They didn’t know what could possibly be wrong with her and didn’t know where else to turn.
The CDC states that roughly 3% of babies out of the 11,000 born every day have a rare genetic disorder that leaves doctors perplexed. Sebastiana was one of the unlucky ones, with both severe neurological damage and developmental delay. It seemed like all was lost until Stephen Kingsmore, the head of Fabric Enterprise, stepped in.
He took a vial of the baby’s blood and sent it to the lab to be sequenced. Fabric Enterprise is a company that uses AI to filter through genomic data using multi-gene panels to narrow in on rare genetic diseases. They used the Fabric ACE method, which rapidly sorted gene variants into a classification engine to be analyzed. In just four days, she was diagnosed with Ohtahara syndrome.
Ohtahara syndrome is an uncommon type of epilepsy characterized by seizures and developmental issues. Most infants don’t make it past age two, but with the help of gene sequencing and AI, Sebastiana was able to fight through it. Finally, the doctors had a clear sense of what was going on in Sebastiana’s head and adjusted her medication accordingly. Three weeks after, she was sent home and got to spend her first Christmas with her family.
Without gene sequencing, the diagnosis would have taken a minimum of six weeks. At that point, the damage to her brain would have been much too great. And if artificial intelligence wasn’t around to speed up the process, there could have been even more dire circumstances. Now, Sebastiana is living happily, all thanks to this revolutionary technology.
People don’t understand how beneficial gene sequencing really is. It is truly the most practical and evolutionary advancement of our lifetime. And especially when paired with AI, there’s no telling what it could do in the future. While the cost of rapid DNA sequencing continues to fall, challenges remain in utilizing this valuable tool to make quick diagnostic decisions. But the main problem with this system is that hardly anybody knows about it or knows the possible benefits of getting a scan.
That’s where Medigene comes in.
A New Type of Opportunity
Medigene is a website where you can get your genome analyzed in a quick and affordable way. There’s no hassle, meaning that you can do it in a timely manner without having to worry about getting your information from anywhere else. Everything is presented in an easy-to-understand format and you’ll never have to worry about overcomplicated results from your doctor again!
It utilizes an artificial intelligence program that allows rapid diagnosis so life-saving interventions can be made as soon as possible, such as immediate treatment and personalized medicine. The sequencing and analysis process is done by an AI system rather than teams of genetic analysts, who typically need to manually shift through gene sequencing to create a proper end result. Missed details on genetic variation result from human error or restricted time, which lowers the ability to treat the patient. Additionally, with AI analysis, treatment plans for patients can be more personalized and effective, taking into account more factors of genetic variation.
By submitting your information to the site, you take the inconvenience out of going to a clinic and getting it sequenced. You just send it in and within a week, you have your data in a clean graphic organizer to look at and analyze as you wish.
The main goal is to make gene sequencing more mainstream in order to allow individuals to have data that they can further utilize for markers of disease. There is a wide range of opportunities, one being the ability to personalize medicine and take a person’s genetic sequence into consideration when tailoring specific treatments.
So What Is Gene Sequencing?
Gene sequencing is the process of organizing and ordering a person’s nucleotide pairs. It is, in and of itself, a pretty complex process. First, a person submits their DNA or multiple types of cells (skin cells, saliva, tissues, etc) to a lab. The DNA then goes through an extraction period.
Step 1: Lysis — Whether it be by mechanical or chemical disruption, the cell membrane and the nucleus are broken open. Dissolving agents, most commonly Proteinase K, are used to free the DNA and get rid of proteins bound to the cell.
Step 2: Precipitation — Once the DNA is stripped away from the cell, it’s still mixed up with the chopped tissues that were originally bound to it. To separate the debris from the DNA, an ion, usually Sodium, will stabilize the DNA molecules and cancel out their charges. Then alcohol is added to slip the DNA out of the other particles, acting like lubrication for the DNA.
Step 3: Purification — Finally, now that the DNA is fully separated, it is rinsed with alcohol again and redissolved in water for easy use and storage.
After the extraction period, it goes through another processing period that we can group into Next-Gen Sequencing (NGS). The DNA sample is sorted into different sized strands and pooled together into multiple adaptor “sequences” that are used to create easier-to-read data sets. Imagine a genetic “library” and that all of the books filling the library are the base pairs being sampled together.
It then goes through a short period of clonal amplification, which is basically making the DNA bigger and even easier to read. It’s then run through another sequence library and the raw computer data is sorted back into base pairs of adenine, thymine, guanine, and cytosine.
I’m practically out of breath!
Yes, I know it’s a lot to take in and you don’t have to understand it all right now. But what you do have to understand is why Medigene is so revolutionary in terms of this new technology.
The Power of AI
Here’s the problem: scientists haven’t implemented this technology into regular hospitals and clinics. Why not?
Well, there’s a few things that we have to keep in mind.
The whole process is pretty expensive, coming in at just less than $1000 per test. That, coupled with the fact that it takes up to 12 weeks to get the data analyzed, makes it extremely inefficient and difficult to get results back in a timely manner.
But there’s another key player that you might have forgotten about that is the sole reason why gene sequencing is progressing so rapidly- artificial intelligence.
There are multiple different types of neural networks. Recurrent neural networks are commonly used for generating text and have been used in gene sequencing before. Multilayer Perceptions solve problems, predict datasets, and classify certain variables into groups. Then comes a convolutional neural network (CNN), which is the most popular and widespread method for analyzing images.
The CNN picks out patterns in entangled layers that receive inputs. Those inputs then go through a series of filters that detect certain patterns. Convolutional networks are simply neural networks that use convolution in place of general matrix multiplication in at least one of their layers. Every layer is made up of a set of neurons, where each layer is fully connected to all of the neurons in the layer before.
Finally, there is a last fully-connected layer (the output layer) that represents the predictions. After it runs through the system, it comes out with more outputs and can tell what certain images are. In this case, the data is unstructured and requires more storage, but the outputs could be basically anything you can think of.
Medigene takes this concept and expands it to gene sequencing.
First comes the training period.
The AI learns by retaining the memory on the features of different nucleotide segments and patterns from images that are labeled. For example, the repeated CTG nucleotide segment that causes myotonic dystrophy can be converted into a short chromatogram that the AI system has to memorize as a repeat that leads to myotonic dystrophy. The AI system would also memorize the features of these recordings with a healthy standard human sequence, like studying for a test. This way, the AI is able to detect differences between the patient sequence and a standard healthy sequence while also being able to identify the type of mutation it spotted. The simulations can be processed, turned into images, and fed back through the system again and again until it creates a practically perfect model to analyze statistics.
Once the artificial intelligence model is trained, it's time to analyze!
Only 2% of our DNA is not junk DNA (DNA that doesn’t indicate anything important). Our AI system analyzes the nucleotide sequences from the 2% of DNA that correlate with protein production in our bodies. Most gene sequence analysis companies are not designed to detect repeated, inverted, inserted, translocated, or deleted DNA. However, these mutations lead to diseases such as fragile X syndrome and Williams syndrome.
A way to quickly analyze and process different segments of DNA is by converting nucleotide sequences to chromatogram images that convolutional networks from AI can analyze very quickly. Limited memory AI that uses convolutional networks — in addition to having the capabilities of purely reactive machines — are also capable of learning from historical data to make decisions. These chromatogram images containing the data of specific nucleotides that make up a DNA sequence (input data) are then filtered and analyzed by the convolutional networks of the AI system.
The different purines and pyrimidines (nitrogen bases) are marked in different colors and converted to images that are processed quickly, efficiently, accurately, and inexpensively by an AI program based on convolutional networks. Different neural layers made of convolutional neurons filter through the sizes (number of pixels) detect the color that corresponds to the purine or pyrimidine and find the distance between their heights on the chromatogram. This process allows observation and processing of huge amounts of data in a much shorter time rather than having teams of medical professionals shift through images, sequences, or graphs.
The AI system detects the details of repeated, inserted, inverted, translocated, or deleted DNA segments that lead to dangerous mutations by comparing differences in the patterns of nucleotides compared to a normal human genome nucleotide sequence in its memory. Due to the speed of AI shifting through the very specifics of a nucleotide sequence, the process of detecting the smallest mutations becomes more accurate while saving so much more time by an automated system without the need for manual analysis by geneticists.
After the AI has identified the mutations, the outputs on the detected mutations are then imputed into the website database where the first person to access the mutation detections is actually the patient or their family members. The data is put into a simple, visual format to make sure the patient and their family know exactly what the results mean. Patient’s have control over their own DNA sequencing results and data. When the doctors and medical professionals have the detection results from the AI system, they can start to immediately deploy appropriate treatment and medication.
So now, we have genetic sequencing and artificial intelligence working together!
The Ins and Outs
The website is very straightforward. It opens up with a few statements about Medigene’s mission and how the process works. You can get more information by filling out the form and a company representative will get back to you. You can make your account by clicking the avatar icon in the top right-hand corner of the screen.
You are then taken to a starting screen where it asks if you’re a patient or a physician.
The Patient Dashboard would present your results and explanations for those results, all in a neat graphic organizer. It shows you the mutated gene and gives a breakdown of the information while also referencing certain symptoms you may have.
The Physician Dashboard presents registered patients’ data, with a list of information once they’ve given the website permission. It also gives any underlying conditions or allergies. It gives the genetic results and treatment options, but also provides them with the best course of action for each patient.
The Future of Biotechnology
So how does this translate to the future?
This website aims to democratize genetic sequencing and make it available to everyone for faster and cheaper results. There’s no need for a geneticist with the use of AI and everyone could finally have sensible ownership over their health data. Rapid diagnosis is a big part of treatment; this enables doctors to perform potentially life-saving interventions as soon as possible. In certain cases, these verifications can save up to 80% of people, billions of lives, that would otherwise have to wait for detection.
Let’s look back at Sebastiana’s case. Ohtahara syndrome currently takes around two weeks to get a diagnosis. An AI automated system could speed up that process by 20%, reaching a consensus in less than 20 hours. Automated systems require 90% less manpower and are extremely important for those suffering from metabolic abnormalities or immunodeficiencies. This system would also cost at least $150 dollars less from typical $200 sequencing results.
Everyone wants to cut down the time that it takes to find a medication that works for them. People also want something simple to understand and an easy way to reach out to doctors and medical professionals for results and other important information.
By utilizing gene sequencing, people will be able to prevent or slow certain diseases and tell when certain factors are arising. There is no need for a group of genetic analysts to manually shift through data and find the mutations or polygenic traits that make up an individual.
Think about your closest family and friends. We all want what’s best for them. In my lifetime, I’ve had people I care about pass away due to a late diagnosis or mistreatment. Many of these diseases, if handled properly and efficiently, could be mitigated and never bother a person again. People have had to pick their battles between living in fear of disease and getting treatments that may not work or make the situation worse.
But now, with the power of Medigene, you don’t have to choose. Just choose “now”.
Get sequenced. Get treated. Get healthy.
Read More:
*When entering our webpage, log in using the email medigene@gmail.com with the password aihackathon.
If you have any questions, email me at vholmes113@gmail.com.
