


In the initial discovery of a drug, it can take cancer researchers up to two years to narrow down 10,000 molecules to find the most effective ones that will safely cross the blood-brain barrier to kill cancer cells. But now, with artificial intelligence, the process can take researchers less than a week.
Panna Sharma, the CEO of oncology drug company Lantern Pharma, says utilizing AI can compress the time it takes to develop novelty drug molecules by 70-80%.
Since 2018, Lantern Pharma has used RADR technology — an AI platform with over 60 billion oncology-specific clinical and preclinical data points, over 130,000 patient records, and 200 machine-learning algorithms that researchers will run different molecules through.
“Its focus is really to compress the timeline of trying to figure out how a molecule works, and, very importantly, which cancer indications to go after, or to design a whole new molecule,” Sharma told the Washington Examiner.
Just last week, the Food and Drug Administration just awarded Lantern Pharma’s LP-184 three rare pediatric-disease designations for treatment in malignant rhabdoid tumors, rhabdomyosarcoma, and hepatoblastoma.
“Usually, that can take half a decade to decade to figure out,” Sharma said. “We’ve done that a matter of a couple of years.”
With less than 200,000 cancer cases, the development of drugs specifically tailored toward combating these specific pediatric cancers have not been explored, primarily due to the cost of development.
Between laboratory research clinical trials, it’s estimated to cost between $1 billion to $2 billion to introduce a new drug to the market — a cost many companies may not want to take on in developing more rare diseases.
Instead, children faced with rare malignant rhabdoid tumors will be given a drug that was created with an adult’s body chemistry in mind.
But with AI technology, researchers will be able to save money and cut down on costs — allowing them to explore more precise treatments for more rare cancers and diseases.
“Imagine now we can get in earlier into their biology and hack it to get rid of these cancers,” Sharma said. “I think you’re going to not only have an era of healthier children post-therapy, but also drugs that are specifically geared toward more early stage and embryonic cancers. That’s very exciting, so that’s a whole new world of biology that is possible because of computation and AI.”
LP-184 will soon be enrolling 50 to 60 patients across a range of solid tumors for its Phase 1A clinical trial.
AI’s incorporation in healthcare and drug development is predicted to increase to $164 billion, a 1000% increase from $13.8 billion in 2022.
The FDA is on track to approve 1,000 AI medical algorithms by the end of this year.
However, concerns continue to exist around AI algorithms.
Xiaoxuan Liu, a clinical researcher based in the United Kingdom at the University of Birmingham, told Nature that results from AI clinical trials can only be extrapolated if the participants are representative of the population the tool will be used in.
“It’s simply a known fact that AI algorithms are very fragile when they are used on data that is different from the data that it was trained on,” Liu said.
Sharma has found that the greatest challenge with AI, especially in testing involving rare diseases, is to see whether there is enough patient data to draw a conclusion.
CLICK HERE TO READ MORE FROM THE WASHINGTON EXAMINER
“Systems are really good at drawing conclusions, if that’s what you ask them to do,” Sharma said. “But if it’s only based on six patients, is that a conclusion that you’re comfortable with, happy with? Is it statistically significant? So I would say the biggest challenge is having enough statistically significant data in some of these ultra-rare brain cancers.”
But just as other fields become more accustomed to AI, learning models and algorithms in drug development will advance and improve over time, continuing to assist companies such as Lantern Pharma as they develop life-saving oncological drugs.