The 137-year-old pharmaceutical and medical device company is going differently in its quest to find new medications.
One of the most significant investments in the healthcare sector is being made by Johnson & Johnson, using artificial intelligence and data science to support its operations.
The 137-year-old manufacturer of pharmaceuticals and medical devices has invested hundreds of millions of dollars in hiring 6,000 data scientists and digital specialists, and they have been tasked with tasks like utilizing robots to search through enormous datasets of health records. The business established a cutting-edge research facility with superior data science last year, close to San Francisco.
Early attempts center on diagnoses; for example, voice recognition technology analyzes speech for early indicators of Alzheimer’s disease, and an algorithm analyzes cardiac testing to identify a dangerous type of high blood pressure much earlier than people can. A virtual reality goggles set is available to educate doctors on procedures such as knee replacements.
Using AI for drug development is a long-term objective that has garnered much attention but has yet to show any indication that it will actually happen.
AI-discovered medications are currently being tested on humans by startup biotech companies. This year, Google unveiled cloud-based AI technologies to help pharmaceutical companies discover novel therapeutics. However, it might take years before authorities approve an AI-discovered medication for commercialization.
Leading figures in the pharmaceutical industry have voiced doubts that AI would ever be able to find new medications more effectively than humans.
With a vast database known as med. AIAI, which can search for patterns to expedite drug development, J&J claims to have an advantage. The dataset contains years’ worth of clinical trial outcomes and “real-world data,” anonymized data gathered from routine patient visits to physicians and hospitals.
“AIAI and data science are going to be the heart of how we are transforming and innovating,” says Najat Khan, chief data science officer and global head of strategy and operations for J&J’s pharmaceutical research unit. “The amount of data is increasing, the algorithms are getting better, the computers are getting better.”
J&J has already used machine learning to create an experimental cancer medication that will begin human trials in 2019.
According to Khan, there are a few differences in J&J’s attempt. Employees in data science are closely involved in the company’s strategic drug research decisions. Tens of thousands of workers have access to the company’s enormous databases, which total more than three petabytes of data in the medical AIAI field. Additionally, it has employed individuals with backgrounds in chemistry, biology, and medication research in addition to data scientists.
With a PhD in organic chemistry, Khan advised pharmaceutical companies on R&D tactics at Boston Consulting Group before joining J&J in 2018. She collaborated with the scientists and was chosen to lead the pharmaceutical R&D operation’s use of data science.
Analysts believe that J&J is among the biggest pharmaceutical companies actively pursuing AI. It was recently placed third out of 50 companies in the Pharma AIAI. The readiness Index by market research firm CBCB Insights examines businesses’ patent applications, investments, deal-making, and other AI-related activities.
Although data-based projects have long been a part of J&J’s expansive business, which employs over 130,000 people and generates $80 billion in yearly global sales, the company’s executives started to adopt a more unified strategy about ten years ago, and they increased spending approximately four years ago.
Five years ago, only a few of the company’s drug-development programs included data science. Today, most of them do. Its research facility in Brisbane, California, near San Francisco, California, integrates data science initiatives with R&D to cure infectious and retinal disorders. Many of J&J’s data workers are dispersed around the company’s several locations, which include Belgium, China, and the United States.
According to Mathai Mammen, who oversaw J&J’s pharmaceutical R&D business from 2018 to early this year and contributed to developing the company’s data-science capabilities, the search for precision medicine was one driving force. Treatments in precision medicine are tailored to a patient’s specific ailment based on genetic or other differences. J&J hopes to learn more about the biological characteristics of disease and how to target treatments that take advantage of those aspects by making intelligent use of its data.
“The patients are more apt to be found in the world and matched with the medicine that matters,” says Mammen, now CEO of the biotech company FogPharma.
In one recent project, J&J scientists led a collaboration of 13 drug companies that analyzed blood samples collected from more than 50,000 people in the UK as part of a national database called UK Biobank. They identified thousands of genetic variants that influence levels of specific blood proteins, about 80% of which weren’t previously known. J&J plans to analyze the dataset using AI and machine learning to help spot patterns. This, in turn, could lead to new drugs or diagnostics that target the gene-protein links to various diseases. In the past, industry scientists would look for such molecular drug targets by scouring academic papers. The AI-enabled approach could spot many more targets more quickly.
The business is also employing artificial intelligence (AI) to analyze digital images of biopsies to find minute variations amongst cancers, leading to the identification of genetic subtypes for certain malignancies. Researchers could use this information to create a medication targeting the genetic subtype.
Collaborations—more than 50 external relationships with data-science startups and others—are a defining feature of J&J’s strategy. “Compared to some of the other life-sciences firms, they seem to be investing more in other companies, startups, and initiatives, in a more venture-feeling sort of way,” says Daniel Faggella, CEO and head of research at Emerj Artificial Intelligence Research, a Boston-based company that carries out market research on corporate use of A.I.
In collaboration with the Mayo Clinic and the Cambridge, Massachusetts-based health technology startup Anumana, one study attempts to expedite the detection of pulmonary hypertension, or elevated lung blood pressure. These days, it may take two years or longer to diagnose the fatal illness.
J&J and its partners gathered six million patient records, all devoid of patient names—encompassing more than eight million ECG readings. Recording the electrical signals in the heart is called an electrocardiogram, or ECG. To train a software algorithm to identify patterns in electrical readings that were present in individuals who were subsequently diagnosed with pulmonary hypertension, they entered the records into the algorithm. According to J&J, the time it takes to diagnose pulmonary hypertension can be shortened by 12 to 18 months when the algorithm is used in conjunction with ECGs.
The algorithm has been designated as a “breakthrough device” by the Food and Drug Administration, a distinction given to items that have the potential to enhance the diagnosis or treatment of critical illnesses. The algorithm has yet to receive FDA approval, which could change in the upcoming year.
