24.03 Liontrust Global Innovation Report - The Rise of AI 04.24 - Flipbook - Page 32
RECURSION
MICHAEL SECORA (PHD), CFO
Michael Secora, PhD, is CFO of Recursion Pharmaceuticals, a biopharmaceutical
company reshaping drug discovery using artificial intelligence and machine
learning. Dr Secora holds a Ph.D. from Princeton University in Applied and
Computational Mathematics and a B.S. in Mathematics and Physics from MIT.
How transformative an impact do you see AI having
on the healthcare industry?
I believe that it is becoming increasingly clear that AI
and ML will inevitably touch various aspects of our lives,
including healthcare. I do not see any part of the value
chain where AI and ML could not have an effect.
I think it is worth highlighting a publication that the Food and
Drug Administration (FDA) released last May, which delves
into the application of AI and ML across the spectrum of drug
discovery and development, starting from target identification to
next-generation manufacturing processes. I believe seeing such a
leading regulatory body outline the applications of AI and ML was
an important moment.
More broadly, there are several sources which frame that the volume
of data generated worldwide is doubling every year or two. Think
about that. Over a short period of time, we are producing more data
than has been generated in human history. How do we begin to make
sense of this vast amount of information? I believe that it is evident
that AI and ML are no longer a luxury, but will become the norm for
parsing, understanding, and finding insights in this data deluge, and I
believe that this trend will be true for healthcare as well.
How do you see the value creation of data and AI
emerging through the healthcare value chain?
Perhaps I will address this question in the context of
what Recursion does. By generating, aggregating,
and integrating vast amounts of relatable biological,
chemical, and patient data, compiled with attention to
standardisation so that every experiment conducted can connect
with every past and future experiment, one adopts a technology
company philosophy within a biotechnology context.
With an approach focused on mapping relatable biological,
chemical, and patient data, one can explore in an unbiased
manner what novel targets could be implicated in a given disease.
In so doing, this process begins to resemble more of a search
problem. For example, given an indication, one could search for
the top 10 potential targets implicated in that disease as well as the
top 10 compounds that could interact with those targets, prioritising
and filtering by target novelty and chemical and clinical tractability.
One can then experimentally validate hypotheses with which one
also improves the overall process, one’s understanding of biological
mechanisms, and the design of chemical matter. Furthermore,
invoking patient-centric data (like we have with Tempus), one can
start to connect cellular-level insights to patient-level understanding,
thus generating causal rationale for drug design.
What is going to be the impact of all of this on us as
healthcare consumers?
For healthcare consumers, I believe that the impact
could unfold in two main ways: increased efficiency
32 - The rise of AI: Technology and Innovation Report
in drug development, which could lead to faster and potentially
cheaper access to therapies; and the discovery of novel treatments
for diseases that were previously untreatable. It is about pushing the
boundaries of known science and opening up new possibilities for
treatment, which is incredibly exciting.
In terms of efficiency in drug development, when thinking about
the process, one should start by surveying a wide area of biology,
and use aggressive science to terminate weaker hypotheses early.
This design choice should help minimise dollar-weighted failure by
focusing resources on the most promising programs and seeking
to reduce costly late-stage failures. Using this approach has led
Recursion to advance programs about twice as fast and for about
half the cost compared to the industry average.
Another critical aspect is novelty. The vast majority of biology
is still unknown. By using AI/ML to differentiate targets and
chemical compounds within large proprietary datasets, one can
then compare these insights to the corpus of scientific literature
and identify opportunities not only to validate known biology and
chemistry but also to uncover unknown relationships. This is what
Recursion is doing which has led to novel relationships that we
have been advancing internally and for our partners. As a scientist
by background, I see this as scientific exploration at scale.
What has got you most excited so far about AI’s
long-term potential in healthcare?
I believe that with greater data analysis, as enabled by
AI/ML, all disease increasingly becomes rare disease
and all medicine increasingly becomes precision medicine. With
greater access to large-scale, relatable data, I believe that we can
start to understand the unique causes of a disease itself, and what
this means on a patient by patient basis.
Take pancreatic cancer for example. It is not just one condition; there
are many forms of it, each with its own nuances. The complexity
is immense, and can involve significantly different genetic or
epigenetic factors. But whether we’re talking about different
types of cancers or other diseases, this level of detail allows us to
understand and address these diseases in a much more tailored
way. With more data, you can aim to solve the precise problem for
the right patient with the right drug at the right time.
As we look to the future, maybe 10 years or so ahead, I believe we
could move from multiomics data to “my-omics” data. The idea could be
to profile each individual across various omics disciplines – phenomics,
genomics, transcriptomics, proteomics, etc. – at a particular stage in
their life, and then provide the precise treatment needed at that time.
As omics technologies become more advanced, widespread, and costefficient, this could allow us to profile ourselves in depth. This could lead
to a situation where healthcare is personalised to the extent of “just in
time diagnosis, just in time treatment”, mirroring the immediacy we see
in consumer product delivery but applied to healthcare.
