A drug developed by combining artificial intelligence with traditional drug development methods has shown early promise in the treatment of some cancer patients in trials.
The results, which suggest the new drug does indeed slow the growth rate of cancerous cells, were obtained by training AI to spot and suggest ways the transformation of a healthy cell into a cancer cell could be reversed.
In trials on 85 cancer patients, the drug led to a 25 per cent reduction in the size of a tumour in one patient.
BPM 31510, a compound made with an enzyme that plays a key role in cellular metabolism, was the first treatment suggested by Berg’s AI. The development of the drug effectively began back in 2009, when the startup bought cancer tissue samples from more than 1,000 cancer patients at various medical schools from across the US. These samples included over 40 types of cancer cells – multiple types of breast cancer, prostate cancer, liver cancer, kidney cancer, lung cancer, brain cancer – along with healthy tissue extracted from the same patients.
Speaking to WIRED in March, Dr Narain said he didn’t want to narrow down his study to just one cancer.
The startup fighting cancer with AI. For now, Berg’s concentrating on making sure the drug is used for treatments that are most likely to be beneficial, in this case, focusing on high-energy diseases like pancreatic cancer.
Microsoft researchers also think that by combining search data from the web, it’s possible to identify people who may have pancreatic cancer before they’ve even been diagnosed.