A future of tailored treatments created by AI. 🩺 Last January, Salesforce AI Research (CRM) launched ProGen, a AI mass capacity language model trained on around 280 million protein sequences to create a protein language model (PLM). ProGen architect Ali Madani PhD said at the time, ‘proteins are responsible for almost all biological processes critical to life’. He said because protein sequences function like a language, the sequences make effective data for machine learning.

Pinning hopes on AI to help ‘discover medicines’. 💊 Madani is now looking at applications for PLMs through his start-up, Profluent — attracting early investors, including OpenAI, Salesforce and Google (GOOGL, GOOG). Profluent trains AI models on vast data sets — using more than 40 billion protein sequences — and can build artificial proteins ‘from scratch’ that are as effective as naturally generated proteins. The company hopes AI applications will accelerate genetic medicine development and protein applications for use ‘from therapeutics to degrading plastic’. 

Speaking to TechCrunch, Madani said Profluent wants to start ‘reversing the drug development paradigm’. That is, starting with the individual patient’s needs, and working backwards to create  ‘custom-fit’ drugs and treatments. AI can also speed up the many years-long research component of a drug’s pathway to FDA approval.

‘[Profluent’s] capability offers humanity a chance to move from accidental discovery to intentional design of our most needed solutions in biology.’ — Ali Madani, TechCrunch

AI healthcare could be worth billions

Covid was a big wake-up call for AI health research. 😷 According to data company Gitnux, the global AI healthcare market could be worth US$50 billion by 2028, saving the US economy as much as US$150 billion every year. 

‘Since 2021, there have been 281 venture capital deals worldwide in AI drug discovery startups, accounting for US$7.7 billion in investment, according to PitchBook.’ — Forbes

Possible benefits of AI technology include:

• Improving cancer treatment outcomes by 30% to 40%
• Halving the time needed for drug clinical trials
• Reducing hospital stays by 50% 
• Saving hospitals US$20 billion in productivity gains each year 
• Automating processes saving up to 20% of clinicians’ time

AI has potential applications in fertility treatment too. 🧬 Globally, around 48.5 million people face infertility issues. Research published in 2021 indicated AI could help clinicians improve treatment for people trying for a baby throughout the fertility process: 

‘AI may aid in selecting the best oocyte and sperm combination as well as predicting embryo quality. Furthermore, AI may assist the clinician in developing an optimal patient-specific treatment regimen to improve IVF success.’ — Does artificial intelligence have a role in the IVF clinic?

These advances could improve in vitro fertilisation (IVF) in the embryo development process by up to 50% and increase the chances of having a live birth by 4%.

Big Tech has invested billions in AI’s biotech future

Slack’s owner Salesforce is not the only tech company applying AI to healthcare. 👩‍⚕️ Today Big Tech is backing AI to the tune of billions, including: 

• Amazon’s (AMZN) machine learning ‘protein folding’ tools on SageMaker, aiming to also accelerate drug discovery
• IBM Research (IBM) is engaged in building software and AI for converting health data for possible healthcare applications with a focus on creating equitable health outcomes. 
• Nvidia’s (NVDA) venture capital arm Nventure, which is investing a large proportion of capital into drug discovery. In 2022 they released open source drug discovery deep learning model MegaMolBART that uses NVIDIA's NeMo-Megatron, which can also be used to predict chemical reactions
• Google DeepMind and their Google AI lab’s AlphaFold protein structure database provides open access to more 200 million protein structure predictions, which is fast with high accuracy  
• Meta’s (META) ESM Metagenomic Atlas delves into metagenomic structures aiming to discover proteins that could help cure diseases, as well as clean up the environment, and create cleaner energy, and is sharing their models and a database of more than 600 million metagenomic structures

Microsoft’s (MSFT) open source protein-generating AI EvoDiff may help create drug delivery methods and therapeutic enzymes

Nvidia’s backing health, from genomics to devices

Nvidia’s fast computing accelerating AI healthcare potential. 🩻 During his keynote address at the 2024 Nvidia AI GTC Conference this March, CEO Jensen Huang announced multiple AI healthcare partnerships. This included with: 

• Johnson & Johnson MedTech (JNJ) to ‘scale AI for surgery’
• GE Healthcare (GEHC) to improve ultrasound imaging

The announcements followed Huang’s January fireside chat with Recursion Pharmaceuticals (RXRX) at the JPMorgan Healthcare Conference. Speaking with Recursion chair Martin Chavez, Huang said:

‘A medical instrument is never going to be the same again. Ultrasound systems, CT scan systems, all kinds of instruments — they’re always going to be a device plus a whole bunch of AIs. The value that will create are going to be incredible.’ — CEO Jensen Huang, JPMorgan Healthcare Conference

Recursion has collaborated with Nvidia for several years, and Nvidia vice president of healthcare Kimberly Powell announced at the conference Recursion as the first hosting partner on Nvidia’s BioNeMo platform. Powell also spoke about other BioNeMo applications, which range from biotech to pharmaceuticals, including with Amgen (AMGN) for drug discovery, as well as Deloitte, Innophore, Insilico Medicine, OneAngstrom and Terray Therapeutics.

Also in March, Ozempic weight-loss drug maker Novo Nordisk (NVO - ADR) based in Denmark said they’re funnelling profits into ‘funding a new supercomputer’ called Gefion, which is powered by Nvidia AI, with a range of health research pilot projects hoped to start at the end of this year.

AI healthcare research is growing at pace

Other companies investing in AI solutions in healthcare include:

• BioXcel Therapeutics (BTAI) uses AI to create new medicines for brain-related health issues. They analyse existing drugs, using their AI platform to find better ways to treat conditions like Alzheimer’s and Parkinson’s. BioXcel has been granted Fast Track Designation by the FDA for their drug BXCL501 — a medicine that helps calm down restlessness or anxiety for people with conditions like schizophrenia or bipolar disorder
• Certara (CERT) uses specialised AI to help scientists with their platform Certara.AI, which has up-to-date data and helps in new drug development.
• Exscientia (EXAI) hopes to use ‘AI to cure disease’ by revolutionising ‘discovery, design and development’ of new drugs. Exscientia announced a $674 million deal with Merck (MRK) hoping to create new drugs in the treatment of cancer, neuroinflammation, caused by infection, traumatic brain injury, toxic substances or autoimmunity as well as Alzheimer’s and Parkinson’s, and immunology. In 2017, Exscientia partnered with Sanofi (SNY - ADR) in a $273 million licensing deal to find small molecule drugs to treat metabolic diseases, such as diabetes, Gaucher’s disease and hemochromatosis. More recently, in 2022, they expanded their collaboration to develop up to 15 small molecule drugs for cancer treatment and immunology. 
• GE Healthcare (GEHC) late last year launched their MyBreastAI Suite for mammograms, designed to help clinicians detect breast cancer using AI looking at breast density
• Iktos is collaborating with Pfizer (PFE) and Merck (MRK) using their products DockAI, Makya, Spaya and Spaya API which can help accelerate drug discovery
• Intuitive Surgical (ISRG) uses their da Vinci robotic surgical systems alongside AI, which helps surgeons understand their patient’s case data
• Medtronic (MDT) is a medical device company that has created an AI Center of Excellence, and they use AI to improve imaging diagnoses for clinicians and doctors using X-rays, MRI scans and CT scans, to speed up AI-enabled healthcare based on device data, and to even address hospital staffing shortages
• Schrodinger (SDGR) creates software for the pharmaceutical industry that uses machine learning to speed up drug discovery

As companies invest billions in AI healthcare to solve the world’s biggest health challenges, AI could be poised to save the world billions of hours in research time and costs.

‍

Like this? 👍 Then you might like: Who stands to gain from Reddit’s IPO?

‍Never miss a Hatch article. Follow the feed on Google News! 📰