The first-ever drug developed by artificial intelligence (AI) has recently entered clinical trials. It will be tested on 60 subjects during the course of the 12-week trial. The drug, currently referred to as INS018_055, will treat chronic lung disease and is being developed by Insilico Medicine. Phase 1 studies of the drug were completed successfully.
An IPF drug
The drug is intended to treat idiopathic pulmonary fibrosis (IPF), a disease in which the lungs scar and breathing becomes increasingly challenging. Around five million people suffer from IPF globally.
A number of therapies can slow the progression of IPF, but none can stop or reverse lung scarring. Patients with chronic scarring have a median survival rate of just three years. The drug will treat IPF – a diseases that makes the lungs scarred and breathing becomes increasingly tough.
Pirfenidone and nintedanib are the current therapies for idiopathic pulmonary fibrosis. Although these medications could offer some respite or lessen the severity of symptoms, they do not undo the harm or stop the development. They also have unfavorable side effects, the most common among them being nausea, diarrhea, weight loss, and appetite reduction.
The majority of those with this serious disease would pass away within two to five years after diagnosis, and there are only very few options for them. Preliminary research suggests that INS018_055 may be able to alleviate some of the drawbacks of existing treatments.
AI Fast-forwards Drug Development
There are four steps in the discovery of any new medicine. In order to treat a disease, researchers must first identify a biological process that is the “target,” typically because it is not operating as it should. Second, they must develop a novel medicine for that target that, like a jigsaw piece, would stop the disease’s growth without endangering the patient. The third phase is to perform investigations, initially on animals, then in human volunteers who are in good health, and lastly on actual patients.
The drug moves on to the fourth and final phase, which is receiving regulatory agency permission for use as a therapy for that ailment if those tests reveal favorable results in treating the disease.
In the conventional method, researchers search public health databases and scientific literature for pathways or genes associated with illnesses in order to identify targets. Artificial intelligence (AI) enables us to analyze enormous amounts of data, discover correlations that human scientists would overlook, and then “imagine” whole new molecules that might be converted into medicines.
In this instance, Insilico utilized AI to both find a novel IPF target and then create a novel chemical that could interact with that target.
Insilico Medicine and INS018_055
Insilico Medicine analyses scientific data from clinical trials and public databases using a program called PandaOmics to find disease-causing targets. Once the target had been identified, researchers entered it into Chemistry42, another tool from Insilico that use generative artificial intelligence to create new compounds.
In essence, Chemistry42 was given the precise specifications needed by the Insilico Medicine experts, and the system then produced a list of potential chemicals and ranked them according to how likely they were to be successful. Because it was the 55th molecule in the sequence and showed the most promising action, the selected molecule, INS018_055, earned its moniker.
The Insilico team is hopeful that the findings of this just-started clinical study will confirm the medication’s safety and effectiveness.
Future of the AI-based Drug Development
The pharmaceutical industry has traditionally been suffering from difficulties launching new drugs. 86% of the recently discovered drugs did not achieve their targets between 2000 and 2015. Traditional medication research in lab settings can take years, but with AI, pharmaceuticals can be conceptualized and prepared for trials in only a few months. For example, Japan’s Takeda Pharmaceutical Co. recently acquired a psoriasis treatment developed in just six months using machine learning to filter molecules to discover the most effective one for a given ailment.
