I’m really excited to tell you guys today how we could use artificial intelligence to discover drugs. One in five Americans is affected by nonalcoholic fatty liver, which has no treatment. It’s usually caused by the accumulation of fats in the liver mostly due to obesity or diabetes.
When the disease is progressive, it leads to devastating complications like liver cancer or liver failure. But can you imagine the drug for treating fatty liver might already exist? For example, there are thousands of drugs that have been developed partially to treat diabetes and obesity.
One of them may have therapeutic effects for fatty liver. But we don’t know. This is just one example. There are thousands of diseases with no treatment. If you could figure out which one of these drugs over here could work for which one of these patients, we can save millions of lives.
In the last decade, because of the availability of genomics data and computer power, AI platforms have been developed to solve this puzzle using genomic data. But genomics alone is not enough. DNA and RNA are really far from what we observe in disease symptoms.
As you may know, DNA will transcribe to RNA, then RNA will translate to proteins. These proteins then interact with thousands of metabolites in the human body.
Metabolites are small molecules, like glucose, cholesterol, all the fats that they were accumulating in the liver of fatty liver patients. The interactions between metabolites and proteins lead to what we observe in disease symptoms.
And they can’t tell you exactly what’s happened in a disease. Now at ReviveMed, we are able to use metabolomics to solve this puzzle.
Until now, metabolomics has been underutilized because metabolites have tremendous diversity in their structures. Each one of them is different from the other one. So it requires several customized processes to identify them in the human body. And that’s why current platforms are only focused to characterize a small number, less than 5% of them.
Our team at MIT developed a proprietary database and AI algorithm to overcome these difficulties. I tell you a little bit about how it works. We have created the most comprehensive database about metabolites, their interactions with proteins, the interaction of proteins to other proteins, the drugs that are targeting these proteins, as well as the association of metabolites with diseases.
We have combined all these data as a gigantic network. It’s a big hairball and it has millions of interactions. And then we do inference on top of that. We then start from the blood or tissues of patients. You think mass spectrometry, we could measure tens of thousands of metabolite masses.
We see, for example, some of them are upregulated or downregulated in a disease. But the problem is that we don’t know what they are because there could be like 10 metabolites with the same mass. For example, if we have this mass of 180, it could be their glucose, or galactose, or fructose. They all have the same molecular mass, but different functions in the human body.
Current platforms — they have to do more experiments to figure this out. What we are doing instead, we are using our AI platform, then we find optimal networks that connect these metabolite masses together. And through these connections, we could figure out, OK, this 180 is, for example, here is glucose. This optimal network further provides us with critical therapeutic insight.
The network itself represents disregulated disease pathways and processes. The proteins in this network could be therapeutic targets. So of those metabolites, there could be an existing drug. Let’s me show you know real case example.
Huntington’s disease has no treatment. We were able to identify novel disease pathways. We are able to identify existing drugs with therapeutic effects for Huntington’s disease. And we were able to identify novel therapeutic targets. This was a work that we did at MIT and published in Nature Methods.
We then formed ReviveMed, a start-up company, to bring this technology to the market and create an impact on people’s lives. By leveraging metabolomics and artificial intelligence, we could discover high clinical efficacy drugs while saving hundreds of millions and years from discovery to the clinic.
Currently, we are focused to discover therapeutics for metabolic diseases by initially focusing on nonalcoholic fatty liver. Because as I mentioned earlier, there are significant needs for patients. Importantly, metabolites play a key role in these diseases, so our technology can create the most impact.
In addition to our internal programs, we are also working with pharmaceutical companies to expedite their direct discovery and development. We have the best team to unlock the value of metabolomics from MIT, BroadInstitute, and Fortune 500 pharma companies.
At ReviveMed, we are passionate about transforming metabolomic data into the right therapeutics for the right patients. Thank you.