Using AI/ML in TechBio
“A computer would deserve to be called intelligent if it could deceive a human into believing that it was human.” By Alan Turing
Hi 👋 everyone and welcome back to another edition of MetaphysicalCells on AI Drug Discovery News.
“By all these lovely tokens, September days are here. With summer’s best of weather and autumn’s best of cheer.” By Helen Hunt Jackson
The Yuan
Last week, three of the featured articles form The Yuan, an Al Community Sees All, Knows All containing in-depth analysis of all the latest trends in the world of AI, were the following:
“Data science reshapes life sciences at the pulsing heart of innovation” by Simone Castello (a writer and digital marketing expert based in Cambridge, UK) about Cambridge’s life science community and the Milner Therapeutics Institute that has built a global alliance of 93 organisations, including 12 pharmaceutical firms in its consortium, as well as affiliated biotech venture companies and research institutions across four continents.
Based at the Milner Therapeutics Institute you can find the 1) Connect: Health Tech (C:HT) that is an inclusive gateway into the University of Cambridge’s life sciences and health tech community for collaborators, companies and investors, and 2) the Functional Genomics Centre (FGC) that is developing novel clustered regularly interspaced short palindromic repeats technologies to understand the biology of cancer, creating biological models and advancing computational approaches to analyse datasets. The Milner Therapeutics Institute provides also space for startups and small and medium enterprises through the Frame Shift Bio-incubator, that complements the Start Codon Accelerator, which provides flexible options for startup companies at different stages in their development.
For example, CardiaTec is a University of Cambridge spinoff led by Prof Namshik Han, Milner Head of Computational Biology, and a world-leading expert in AI for drug discovery, as well as Raphael Peralta and Thelma Zablocki, alumni of the MPhil Bioscience Enterprise. CardiaTec is accelerating its impact by applying AI on large scale multi-omic datasets (i.e. genome, proteome, transcriptome, epigenome and microbiome data) to develop innovative cardiovascular disease drug targets.
Last year, CardiaTec Bioscience secured a £1.4million pre-seed investment (led by Laidlaw Scholars Ventures and APEX Ventures with participation from Crista Galli Ventures, o2h ventures and Cambridge Enterprise) leveraging AI to make sense of large-scale multiomic cardiovascular data, as opposed to conventional singular omic analysis.
The second article “Pharma firms must abide by 10 articles to avoid foundering in a sea of data” by Ganes Kesari (an entrepreneur, AI thought leader, author, adjunct professor, and TEDx speaker) is about 10 critical elements to examine in order to establish a corporate data strategy and realise business value, such as:
Business goals: Consider for example the business goals of Janssen Pharmaceuticals, a part of Johnson & Johnson (J&J): “We aspire to transform lives by bringing lifesaving and life-changing solutions to people who need them. We’re committed to providing safe and effective medicines as well as the services and support that contribute to healthy outcomes.”
D&A (data and analytics) vision: J&J is once again a good example of how this can be orchestrated. As Dr Najat Khan, chief data science officer and global head of Strategy & Operations for Research & Development at Janssen, puts it: “At Janssen, we are applying data science end-to-end across our portfolio by focusing on finding solutions to big questions that can advance our impact on patients.”
Key stakeholders: While 82.6% of organisations have appointed chief data or analytics officers, fewer than half these leaders feel that their teams effectively contribute value to their organisations. Therefore, the careful selection and engagement of key stakeholders and team members is vital.
Strategic initiatives: A pharma firm that aspires to improve its customer experience might plan strategic initiatives around customer targeting and personalised engagement. It may design patient support programs to recommend AI-driven interventions to help support specialists’ interaction with patients and improve patient conversion through targeted outreach backed by analytical insights.
Measures of success: Take the case of J&J Medtech, which aims to deliver patient care with smarter, less invasive, and more personalised digital surgery solutions. Its Advance Case Management platform is a digitally integrated system that simplifies pre-surgery processes with case schedules and patient data. With AI-driven predictive insights, the system has been found to reduce its inventory footprint by 50% and deliver a reduction in the time spent on critical operating room procedures by over four minutes.
Sources of funding: In the world of data, without access to funds, even the best-laid plans are almost certainly doomed. Implementing a robust data strategy requires resources, skilled people, technology infrastructure, or investments to manage change processes. So far, those who either allocate no budget for data initiatives (28%) or only sporadically fund these initiatives via their information technology budget (20%) reportedly together account for nearly half of all organisations.
Leverage Top enablers: CDOs (chief data officers) will have to identifying within their organisations factors that can come to their aid if need be. If a C-suite executive is a strong supporter, the CDO must leverage his/her presence to choose strategic initiatives, drive the programs, and story-tell the message to the entire organisation.
Top challenges: Business leaders must anticipate organisational challenges when executing their data strategy. This can help them devise proactive strategies for mitigation. Unfortunately, this does not always occur in practice since 71% of data leaders lack confidence that their company’s leadership sees a link between investing in D&A and staying competitive.
Governance plan: Organisations must establish mechanisms to track, review, and validate outcomes from their D&A initiatives, and identify ownership and an ongoing cadence to review progress and performance at regular intervals. For something like a cross-functional steering committee consisting of leaders from the executive council, business units, and technology teams, it must take ownership of the company’s data strategy and monitor performance throughout the journey, which will greatly increase the chances that D&A initiatives will succeed.
Operational capabilities: As part of their data and analytics strategy, organisations should plan to invest in operational capabilities across people, processes, and technology. They should find out how to onboard and empower users, how to rewire their business processes to integrate their data initiatives, and how their technology strategies can scale their analytics efforts. So far, only 34% of Fortune 500 companies appear to be strategic about their technology investments, while just 2% have AI systems in widespread production.
Their third featured article “AI-aided diagnosis, treatment bring a ray of hope to ADHD sufferers” by Disha Ganguli (a writer, analyst, educator and child psychologist) is about attention deficit hyperactivity disorder (ADHD) and how AI can enhance its management only when algorithms are accurately and suitably trained.
In fact, if AI is used properly can certainly play a part in eventually curing ADHD. In diagnostic assistance, AI algorithms are adept at analysing a range of data such as behavioural patterns, medical history, and cognitive assessments, which helps psychologists and psychiatrists diagnose ADHD more accurately. This also helps in digital therapeutics, with AI-powered applications assisting patients with cognitive training, exercises, behaviour tracking, and reminders to take medications.
AI in Drug Discovery Market Size & Share to Surpass $ 7.1 Billion by 2030
By Vantage Market Research
The Top Report Findings Are:
AI-driven drug discovery significantly reduces research time, leading to quicker development cycles.
Integration of AI enhances hit identification accuracy and reduces false positives in virtual screening.
Predictive analytics improve the success rate of clinical trials by identifying potential risks and optimising trial parameters.
Top NVIDIA Blogs
Last week, the featured articles and podcasts form NVIDIA on AI pharma and AI drug discovery, were the following:
AI-Fueled Productivity: Generative AI Opens New Era of Efficiency Across Industries by Cliff Edwards (part of the Enterprise Communications team at NVIDIA).
In this blog you will learn about how Amgen using NVIDIA’s BioNeMo models has slashed the time it takes to customise models for molecule screening and optimisation, from three months to just a few weeks. NVIDIA BioNeMo™ is a cloud service for generative AI in drug discovery, allowing researchers to quickly customise and deploy domain-specific, state-of-the-art generative and predictive biomolecular AI models at scale, to rapidly generate the structure and function of proteins and biomolecules, accelerating the creation of new drug candidates.
Another example, comes from Recursion a leading clinical stage TechBio that announced this year a $50 million investment by NVIDIA, which was executed as a private investment in public equity. Recursion also announced plans to accelerate development of its AI foundation models for biology and chemistry, which, in collaboration with NVIDIA, it intends to optimise and distribute to biotechnology companies using NVIDIA cloud services (Recursion Announces Collaboration and $50 Million Investment from NVIDIA to Accelerate Groundbreaking Foundation Models in AI-Enabled Drug Discovery).
Quicker Cures: How Insilico Medicine Uses Generative AI to Accelerate Drug Discovery by Renee Yao (that leads global healthcare AI startups at NVIDIA).
Insilico Medicine, a premier member of NVIDIA Inception, is entering Phase 2 clinical trials with a drug candidate discovered using its AI platform. NVIDIA Inception, is a free program that provides cutting-edge startups with technical training, go-to-market support and AI platform guidance. Insilico uses NVIDIA Tensor Core GPUs in its generative AI drug design engine, Chemistry42, to generate novel molecular structures — and was one of the first adopters of an early precursor to NVIDIA DGX systems in 2015.
“This first drug candidate that’s going to Phase 2 is a true highlight of our end-to-end approach to bridge biology and chemistry with deep learning. This is a significant milestone not only for us, but for everyone in the field of AI-accelerated drug discovery.”
Alex Zhavoronkov, CEO of Insilico Medicine.
Matice Founder and Harvard Professor Jessica Whited on Harnessing Regenerative Species — and AI — for Medical Breakthroughs, by Brian Caulfield (NVIDIA's chief blogger).
On the latest episode of NVIDIA’s AI Podcast, Noah Kravtiz spoke with Jessica Whited that is a regenerative biologist at Harvard University, and co-founder of Matice Biosciences, a company using AI to study the regeneration of tissues in animals known as super-regenerators such as salamanders and planarians.
Introducing AWS HealthImaging — purpose-built for medical imaging at scale by Tehsin Syed and Andy Schuetz at AWS HealthImaging.
AWS HealthImaging, is a purpose-built service that helps builders develop cloud-native applications that store, analyse, and share medical imaging data at petabyte-scale. HealthImaging ingests data in the DICOM P10 format and provides APIs for low-latency retrieval, and purpose-built storage.
HealthImaging is integrated with Amazon SageMaker for ML, giving you access to GPU accelerated computing. Also NVIDIA is investing in hardware accelerated tools and open-source frameworks that work seamlessly with HealthImaging to advance algorithm development and AI adoption across medical imaging.
For example, the MONAI Label reduces the time and effort of clinicians to annotate new datasets, while the underlying AI-based labelling apps continuously learn and improve:
“MONAI, co-founded and accelerated by NVIDIA, is a domain-specific, medical imaging AI framework that speeds the translation of research breakthroughs and AI applications to clinical impact.
With the integration of MONAI and AWS HealthImaging, medical images can be viewed, processed, and segmented in near real time — optimising physician workflows, enhancing patient experiences, and helping hospitals improve efficiencies.”
Prerna Dogra, Global Lead for Healthcare AI Products, NVIDIA
Latest studies on AI/ML tools:
📌 Stanford Medicine developed an algorithm leveraging single-cell RNA-seq and spatial transcriptomics data, to predict transcriptional subtypes of glioblastoma cells from histology images, in order to determine a tumour’s aggressiveness, its genetic makeup and effectiveness of surgery.
📌 International researchers trained a ML model to identify autism spectrum disorders with 95% accuracy, on multiple datasets for four age groups.
📌 UC Berkeley researchers recorded electrical activity from areas of the brain as patients listened to the Pink Floyd song, “Another Brick in the Wall, Part 1”, and by using an AI software they were able to reconstruct the song from the brain recordings. This is the first time a song has been reconstructed from intracranial electroencephalography recordings.
📌 MIT researchers combined deep learning and physics to fix motion-corrupted MRI scans.
Let’s go to Asia now.
Asia
Wonder drugs in the AI age: The Asian advantage by Praseeda Nair
Huawei Cloud is adopting a long-term investment strategy in AI-assisted drug design with the The Huawei Cloud Pangu Drug Molecule Model (named after a figure in Chinese mythology), developed with the Shanghai Institute of Materia Medica, to help pharmaceutical companies build small molecule drugs, using data from over 1.7 billion compound.
“AI could effectively function as a virtual chemist, helping researchers design and identify novel molecules that are likely to interact with drug targets”.
Dr. Qiao Nan, Head of Huawei Cloud EI Health.
On the back of this success, Huawei launched a unique AI-assisted commercial pharmaceutical SaaS platform in China to help companies reduce the costs of trial and error, while accelerating the discovery of lead compounds from several years to just one month. The SaaS platform is slated to expand internationally, starting with APAC and the Middle East.
Currently, China is leading the global AI industry, housing over 60% of big data experts across all sectors, and in particular, according to GlobalData’s Contract Service Provider database Merck currently owns three facilities in Singapore, and Novartis and GSK each own two facilities.
Moreover, last May Japan’s Takeda Pharmaceutical acquired the US-based AI startup Nimbus Therapeutics for $4 billion, a company using AI and ML algorithms to develop a compound to treat psoriasis. The experimental drug has already passed the first two phases of clinical trials.
MORE NEWS
👉 The 16z-Backed AI Biotech Startup Genesis Therapeutics raised $200m for its AI drug discovery engine.
Genesis was set up around a physics-based AI platform the Genesis Exploration of Molecular Space (GEMS), which integrates deep learning and molecular simulations for property prediction, and language models for molecular generation.
👉 Mendaera, a healthcare robotics company, raised a $24M Series A funding led by Lux Capital with participation from Founders Fund and former U.S. Senator Bob Kerrey. Mendaera provides a platform that combines real-time imaging, robotics and AI to enable precise and consistent intervention. In addition, the company is building an ecosystem of imaging and instrumentation partners to expedite minimally invasive care.
An Overview of Google’s AI Product Strategy
As PaLM (Pathways Language Model) matures, Google's ceiling in Generative A.I. is very high for useful products. By
The PaLM set consists of:
PaLM v2, the Transformer - based, generalist model, built to cover a broad range of topics and languages
Med - PaLM, fine-tuned with medical data for healthcare use cases
PaLM - E, embodied multimodal language model for computer vision
Audio PaLM for speech-to-speech translation
Sec-PaLM, a specialised security LLM which powers the Google Cloud Security AI Workbench
Codey, built for software coding assistance, with support for 14 languages, including Python, Java, Javascript and Rust and accessible through Vertex AI
👉 Check the most active investors in the health tech (i.e. standout investors in the health tech sector are OrbiMed and SOSV) and in digital therapeutics (i.e. JAZZ Venture Partners and SOSV) in the listing: The top VC investors in healthtech by Sabine Müller from dealroom.
👉 The AI-powered care facility automation platform care.ai announced that it is partnering with Samsung to integrate its Smart Care Facility Platform into the tech giant's displays for use by health systems (Care.ai's AI-Powered Platform to be Integrated into Samsung's Displays).
👉 The South Korean medical AI company Lunit is helping raise Sweden's cancer screening capability by partnering with one of the country's biggest private healthcare providers (Lunit's AI Solutions: Improving Cancer Detection in Sweden and Healthcare in Military).
👉 Huma Therapeutics, a leading global digital health company, announced this summer that it has received FDA Class II clearance for its disease-agnostic Software as a Medical Device (SaMD) platform (Huma's FDA-Cleared Platform: Personalized Health Data Management Made Easy).
Until next time 🍃,
P.S.
📌 Allow patents on AI-generated inventions — for the good of science by Ryan Abbott a professor of law and health sciences at the School of Law, University of Surrey, Guildford, UK, and adjunct assistant professor of medicine at the David Geffen School of Medicine, University of California, Los Angeles, USA.
Are AI-generated inventions patentable?
“The Artificial Inventor Project has laid bare a jurisdictional split over how AI-generated inventions are treated. Patents have been denied in the United States, Australia and Taiwan, but South Africa issued one that has an AI listed as the inventor and its owner as the patent owner, as Saudi Arabia is expected to do. The European Patent Office has noted that although an AI cannot be named as an inventor, nothing prevents its user or owner from listing themselves instead and disclosing that an invention is AI-generated. Germany’s intermediate federal court has ruled similarly.”
Ryan Abbott
📌 Zagreb Art Singularity, is presenting this month its first exhibition of AI-generated images in cooperation with Global AI Ethics Institute (the only global think tank addressing ethics applied to AI through cultural lenses co-founded and co-directed by Aco Momcilovic).
The VIP Opening of the ZAS Exhibition will be on the 8th of September at the Hotel Zonar Zagreb, Croatia.
The Global AI Ethics Institute is working on different projects right now to raise awareness on the importance of cultures in the ethical appraisal of AI systems such as: Cultural Perspectives On Ethics Applied To AI and Alliance For Responsible AI.