Funding Computational Neuroscience

PART 2: The 68th Lindau Nobel Laureate Meeting @IKUNetwork

The purpose of interviewing both young scientists and Nobel Laureates from the 68th Nobel Laureate Meeting was to speak directly to the source and familiarize ourselves with the experience as a medical scientist. In the first podcast (coming soon) we’ll share these interviews as well as our own perspective, shining a light on the current/future state of biotech and how both ultimately rollup to human longevity.

As the week progressed with more panels and lectures discussing the degradation of the quality of science publishing and dire state of funding, the struggle in receiving grants and funding for research became much more apparent. Even at the high level of Nobel Laureate invitees. The narrative somewhat organically developed throughout the interviews as these young scientists had a lot to say.

Biotech & Funding

For context, ~260 biotechnology products were approved for over 230 indications in the past 30 years, with global sales of these products exceeding $175 billion in 2013 helping sustain ~4,600 biotech companies worldwide. The development in biotech also aids in the evolution in government activity, business development, and patient care — the industry has serious peripheral implications. While these numbers may seem impressive, they only tell a small part of the story …

Ole Kohler-Forsberg @kohler_ole, an MD PhD Neuroscientist at Aaruhus University, shares his perspective on biotech funding as it pertains to his research:

Roughly 40% goes to salary, and then the other part goes to everything around the technology, then of course you need to establish a pipeline for recruiting patients, and then the actual equipment for doing the interventions like lumber puncture, blood tests and all that…but most of the money goes towards the analysis of everything.

Some analysis we do is quite basic and…cheaper. But a lot things like specific antibodies, infectious agents, or some very specific cells…they are quite difficult tests…And if we go into metabolomics, transcriptomics, genetic aspects of it…they require time and because it is very new technology and only very few labs can do these analyses.

Grants play a huge role because of salary, patients, etc. But things costs and the thing with mental disorders you are testing very broad from the beginning, kind of hypothesis generating, so we need to test a broad range of inflammatory markers, immunological markers, both in the blood and in the cerebrofluid analysis. Funding for all kinds of tests and equipment…funding plays a huge role.

With this one subsect in bio R&D you can see how infrastructure and capital are both required to advance research in the field. Grants help to a certain extent…

Grants are essentially the equivalent of charitable donations given to scientists by governments and non-profits — where economic incentive is lacking. Writing a grant application is a daunting task in itself, often having to adhere to strict administrative paperwork, taking brilliant scientists away from what they do best — science. Upon submission of a grant application, only ~10% of the total application pools receive funding. The U.S. National Institute of Health (“NIH”) is the biggest funder of bio R&D grants, but NIH funding has been generally decreasing over the last 10 years. The other option is working with industry but they have their own initiatives — usually requiring patent positions to dictate the scientific experiments as opposed to science dictating patent positions. Traditional models are starting to change with the advancement of computational biology, especially with respect to the brain.

Computational Neuroscientist, Louis-David Lord, a PhD at the University of Oxford, discusses tech in the context of funding his rapidly advancing field:

There’s a couple of different ways to talk about the technology in Computational Neuro. On the one hand you have to think about the technologies that provide us with data. And that is neuro imaging equipment — scanners, MRI’s, magnetoencephalography… the actual data collection tools. The other aspects..is the computer power. You need servers, and you also need software, algorithms. You need all groups that are capable of processing the neuro-imaging data, filtering it, and correcting for motion etc. And then you also need to to make sense of the data in a way that is physiologically relevant and interesting and actually helps you understand the brain, helps you understand diseases, and that to me is the exciting part. Once I have these this data, what do I do with them and which approaches from mathematics or computer science can I use to help understand how the brain works?

Computers obviously play a huge role in Dr. Lord’s work, as well as actual medical equipment like a state-of-the-art 3 Tesla MRI machine. How much do machines like this cost? Roughly $3 million. So not only do you have expensive medical devices required in collecting data, you have Machine Learning “ML” computational science, for example, that incorporates numerous technologies processing oceans of data with ever improving analytical models. The good thing is that with the emergence of various tools, libraries and frameworks for building ML-based software, the ML technology is becoming more available. In fact many of the ML products being developed today cost must less when compared to the early days when Google, Microsoft, etc. introduced the technology. Now a product release can cost somewhere between $100k-$300k. But with any new state-of-the art tech, it always costs, and always will. Especially if it works well.

More on Computational Neuroscience, especially in the context of AI — a super exciting field where essentially enormous amounts of data can be processed to help us better understand, for example, how AI can imitate the brain. Here’s a rundown on specific research in the field from Patrick Malone @patricksmalone, MD PhD candidate at Georgetown University in Computational Neuroscience:

There are some important differences between what these AI algorithms are doing and what what the brain is actually doing so we want to construct an example why there might be some important differences. A lot of times these deep learning architectures, these AI algorithms are trained on very massive data sets of images so you want to train a network to recognize pictures of cats versus dogs for example. You can train the network to get to superhuman performance but then you can also do this interesting experiment where you just change a single pixel within an image of a cat. And if I showed you, as a human, the changed picture versus the original picture you wouldn’t even notice the difference between the two. But these AI algorithms sometimes will just completely crash and burn with that task. So the point is that there’s something important there that’s different about the way the AI algorithm is doing vision versus the way the human brain is doing vision. And I think a lot of AI researchers are kind of realizing now that there’s a lot to be learned from the brain if AI is going to kind of take their performance and their applications in the real world the next the next step. It is certainly been in the last five to 10 years powered by GPS use graphical processing units from companies like Nvidia and research from places like Facebook and Google. They’ve made huge strides. They have huge divisions now dedicated towards AI research.

Beyond data collection tools is the ever evolving techniques and technology used to make sense of the data. Case in point with Patrick’s work. The more data we collect while in parallel the methods developed to analyze spit out negative/positive results evolve, at higher frequency and reduced time, the more equipped the field is with robust conclusions, and the faster the rate of improving data processing techniques continue.

With any new research or technology of importance to the future of humanity, time and resources are required. And with respect to the time it takes for new discoveries and biotechnologies, Gürkan Mollaoğlu, a PhD in Lung Cancer and Tumor Immunology at the University of Utah, summed up the future of R&D quite well:

…the power of computers and AI focus efforts on more promising candidates and approaches, bypassing years of research it typically takes to reach the same result. AI can help digest knowledge much faster than humans. And we’re only at the beginning of this intersection between AI, machine learning, and biotech…

Aligned economic incentives of stakeholders combined with the intersection Dr. Mollaoğlu describes will likely produce the most efficient outcome for pushing science and medicine to boundaries yet unseen.

FYI

Metabolomics: Metabolomics is the large-scale study of small molecules, commonly known as metabolites, within cells, biofluids, tissues or organisms. Collectively, these small molecules and their interactions within a biological system are known as the metabolome.

Transcriptomics: Transcriptomics is the study of the transcriptome — the complete set of RNA transcripts that are produced by the genome, under specific circumstances or in a specific cell — using high-throughput methods, such as microarray analysis. Comparison of transcriptomes allows the identification of genes that are differentially expressed in distinct cell populations, or in response to different treatments.

Young Scientists Bio’s

Ole Kohler-Forsberg — MD Psychiatry & Psychoneuroimmunology at Aaruhus University. Supported by the Danish Council for Independent Research and co-funded by the Foundation Lindau Nobel Laureate Meetings

Research Motivation: Better understand the interplay between the central nervous system and the immune system and hot eh immune system may contribute to the etiology of severe mental disorders, such as schizophrenia or depression and whether this may help in more personalized treatment.

Louis-David Lord — PhD Computational Neuroscience at University of Oxford. Supported by Mars, Inc.

Research Motivation: My motivation for science stems from a deep fascination with the natural world, a desire to understand its intricacies beyond the surface, and the challenge of translating scientific discoveries into life changing applications.

Patrick Malone — MD/PhD in Computational and Cognitive Neuroscience at Georgetown University. Supported by the Boehringer Ingelheim Stiftung

Research Motivation: My scientific motivation comes from a fascination with the brains and how it works. I am driven top discover theories of brain function through the use of computational methods, and how to use these discoveries to find new treatments and cures for neurological and psychiatric disease.

Gürkan Mollaoğlu — PhD in Lung Cancer & Tumor Immunology at the University of Utah. Fellow of the Bayer Science & Education Foundation

Research Motivation: As a young scientist, I believe that increasingly complex issues of our civilization can be be tackled by logical and scientific approaches. I aspire to be a groundbreaking researcher, science advocate, and among the visionary leaders of the next next generation.