Over the past 6 months, we have been on a journey introducing the huge medical opportunities of Artificial Intelligence (AI) in personalized medicine.
The future is quite clear – AI-powered applications are "game changers". This is especially true for the healthcare space – where sooner than later we will be able to completely personalize the diagnosis, treatment and care management for each and every patient on earth!
However, before we embark on the next leg of our trip to this futuristic healthcare space, let us summarize what we have learned so far.
Data is now everywhere – bringing huge opportunities to the medical world
We now live in a world where data is simply everywhere. In fact, by 2020 the digital universe is estimated to contain nearly as many digital bits as there are stars in the universe. And this is doubling in size every two years!
How come? We are now generating and accessing data in everything we do – online searching, shopping, travelling, blogging, eating, reading, posting, travelling, exercising – the list is never ending. All our data is now stored somewhere up there, ideally a secure cloud, just waiting for AI-applications to analyze it and gleam new insights to develop groundbreaking solutions that will change our everyday life.
Just like in all other industries – we are also massively generating data in the healthcare space. The explosion in wearables, medical devices and IoT means we can now benefit from previously unattainable Big Data. This real-world information includes datasets from electronic health records, registries, hospital records, health insurance data alongside biobank, genomic, digital phenotyping information and new brain mapping tools.
Advances in computer science and AI, enable us to analyze all this structured and unstructured “Big Data” to discover new hidden secrets with huge value. Scientists and doctors now have a constant source of Big Data for ongoing discovery and analysis that simply wasn’t available before!
We are now able to use these new AI technologies to analyze and extract insights that will empower our doctors and healthcare practitioners, to personalize and deliver life-changing services for patients.
We welcome the AI-Powered Personalized Medicine new age.
We can unlock the brain’s mysteries
In my first blog, “A New Era in Brain Research: Applying Big Data approaches and Machine Learning algorithms”, I aimed to explain the challenges in researching the most complex organ in the body, the brain. We discovered why it is crucial for us to comprehend the wider impact of the environment on gene expression so we can understand the background disorder of each patient for effective treatment delivery. We also discussed how effective targeted therapy based on gene expression biomarkers must understand the multifaceted and opposing effects of gene expression in distinct brain regions.
My blog finished by emphasizing the medical opportunity we are now given from the unprecedented availability of genetic, clinical and environmental Big Data – we are entering into a new era of AI-powered brain research! Whereby using advanced methodologies, like AI, we can deeply analyze all this new Big Data, to better understand the brain’s complexity and enable transformative innovation in the treatment of psychiatric disorders. We can empower our scientists and doctors to improve medical recommendations and personalize treatment.
We can map the brain’s genetic complexity
In my next blog, I wanted to explain the complex journey from gene to mental health behavior “Cracking the genetic mysteries in the mental health modern space”. We discussed how AI algorithms, like GPS tools, may provide a route to map the brain’s genetic complexity to develop life-changing personalized medicine.
In many health conditions, we now know if someone has an increased likelihood of developing a particular disease based on a person's genetic makeup. For example, in the genetic disorder Fragile X, the mutation of the FMR 1 gene on the X chromosome is directly correlated with mild to moderate intellectual disability. This means we can better diagnose and develop treatments that target this genetic disorder. However, in the case of many psychiatric disorders, no single gene has been found – the story is much more complex...
In this blog, I aimed to explain our body’s genetic intricacies. How our behaviors are the result of complex, often non-linear, gene-gene and gene-environment interactions, combined with random processes associated with the interconnection of millions of neurons and synapses.
The role of brain-derived neurotrophic factor (BDNF) in depression[i], which I researched during my PhD, provided an excellent example of these genetic complexities. It is this complexity that has made progress into the genetics of mental illness exceedingly difficult.
AI provides us with new tools to make sense of these genetic complexities. This is exactly where Taliaz comes in. Based on learnings from my 20 years of scientific brain research, I started the Taliaz company, with one clear vision: To integrate neuro-scientific-genetic models with AI technologies, by innovating usable tools that will empower our doctors to deliver truly personalized treatments for patients.
Our flagship innovation, called Predictix, aims to do just this by analyzing each and every patients’ genetic makeup, clinical and demographic information. It then correlates this with large datasets from the largest ever study on depression therapies (STAR*D), to provide a treatment prediction report. Each personalized report ranks which antidepressant drug is statistically the most effective to each patient’s unique genetic makeup and health record. By using Predictix, our doctors can make informed therapy choices, before prescribing antidepressants, to deliver the promise of personalized medicine.
It is by understanding the genetic landscape along with wider environmental data, that we can select the important features for our AI model.
When building our AI model, we were able to analyze tens of thousands of different gene combinations alongside demographic and clinical Big Data variables. Our AI algorithm, Predictix, was then able to discover and reduce down the number of features to those that matter. We now had an AI algorithm that could help clinicians choose the right antidepressant for their patients sooner.
Unlike other pharmacogenetic tests, Predictix looks much further than one's genetic makeup, to incorporate our demographic and clinical data. So why is this important?
Understanding the complex environmental-genetic journey
In my next blog “Personalizing medicine: Using AI to analyze combinations of genetic and environmental factors”, I aimed to show how we must use AI to go beyond the complex genetic world and incorporate the much larger environmental picture – to personalize treatment to our life experience!
We know that the search for a single gene responsible for many psychiatric disorders is unrealistic. Instead, we must assume they are highly complex disorders in which multiple gene variants, with different levels of influence, are involved.
Though treatment is probably linked with a patient’s genetic background, we cannot forget the environmental impact on our genetic journey.
We know that some people with a genetic predisposition for mental illness will never get the disease while others will, even within the same family. We also know that the timing of these environmental factors can have a profound effect on our health outcomes. For example, the impact of poverty, inappropriate care or violence in our environment during the sensitive periods of early childhood, may route children on a course to emotional, physical and mental health problems. The prevailing theory is that some people are genetically predisposed to mental illness, which are provoked by environmental factors.
Clearly, to personalize medicine and predict which treatment is the most effective for each individual, we must understand our genetic makeup in combination with the environmental factors influencing our genetic journey.
We can change how clinical research is delivered
In my next blog, “The future of clinical research: AI personalized, real-world studies”, I wanted to show the potential impact of AI and real-world data in enabling a new approach to clinical research.
How we no longer have to approach clinical research with a traditional specific, focused question – for example, “Is this drug safe?” The days of running research, with a defined protocol on a select group of patients with defined characteristics and health status, are over.
This blog stresses that the advent of real-world data (all these new databases at our finger-tips) enables a new approach to clinical research. For example, we can ask many different questions in an ongoing way that are constantly fine-tuned and optimized as we collect more data!
We can get trials started and completed in record time that involve and analyze thousands or even millions of patients! We can improve the quality, accuracy and repeatability of studies as we have more patients and datasets. And we can do this all in an uncontrolled, real-life setting at much less cost than ever before!
Real-word data gives us the capabilities to deliver ever-evolving clinical studies that by using AI can capture and analyze billions of data points to collect “real-life” insights. Most importantly, it has the ability to adapt, learn and personalize study treatments to individual patient experiences – in my opinion the “holy grail” for clinical research.
For Taliaz, we are hard at work exploring ways we can deliver new innovative solutions in this changing field. For example, Predictix could be harnessed to help pharmaceutical companies and other drug development companies design trials with those patients who are most likely to respond to treatment, providing huge cost savings.
We can personalize antidepressant treatment
My final blog in my series to date, “The future of depression therapy: Understanding the role of genetics” returned to my PhD and where it all started for Taliaz – with the world’s leading cause of disability – depression!
In this blog, I really wanted to explore the role of genetics in depression. Do genes play a role in depression? What about our environmental factors? As depression is a complex behavior, how can we analyze the vast amounts of available healthcare information to help us understand it, and help doctors better treat it?
This blog went much deeper into the hereditary nature of depression. We showed that there should be specific genetic factors involved in depression that researchers can identify to help better target and improve treatment. We then explored some of the challenges to identify the “depression gene”. Highlighting how researchers believe the failure of the Genome-Wide Association Studies (GWAS) in the last decade to identify significant genetic sites is that depression involves a combination of genes.
Furthermore, the lack of large sample sizes has meant they did not have the adequate power to detect genetic variants of small effect. The recent success of the trial by Pfizer with 23andme demonstrates this point. By using much larger datasets (the trial involved over 300,000 people through 23andMe’s massive genetic database), Pfizer was able to link 15 genome sites to depression using the traditional genome-wide association approach.
Supporting the underlying theme throughout all my articles – we finish with the important role of the environment in depression. Touching once more on McGowan et al[ii] (2009)’s important article in which childhood abuse may lead to epigenetic changes that may increase the behavioral risk of suicide – directly linking how the environment may impact gene expression and our resulting behavior.
Clearly, to truly deliver the promise of personalized medicine, we must understand the intricacies between the complex genetic and environmental landscape. With clever modelling, AI can sift through this vast ocean of data, pulling out and connecting the dots that matter to help us quickly predict the right treatment for each patient.
In the case of depression, with 65% of patients failing to achieve remission[iii] following their first time antidepressant treatment, providing tools that can empower doctors to personalize antidepressant prescriptions will be life saving!
I hope that I have given you a good overview of my different articles and the journey so far. If you have not had a chance to read all my blogs in depth, I recommend clicking through to gain a more comprehensive view of each topic.
In my next blog series, we will go much deeper into subjects to reveal how AI can help us better understand the brain – learning and behavior, brain networks, function and behavioral complexity – to transform treatment of mental illness. We will also discuss the challenges and opportunities of personalized medicine approaches in clinical practice.
Looking forward to it! Until next time!
[i] Taliaz D., Stall N, Dar DE, Zangen A. Knockdown of brain-derived neurotrophic factor in specific brain sites precipitates behaviors associated with depression and reduces neurogenesis. Mol Psychiatry. 2010 Jan;15(1):80-92.
[ii] McGowan PO et al. (2009) Epigenetic regulation of the glucocorticoid receptor in human brain associates with childhood abuse. Nature Neuroscience 12:342–348.
[iii] Khalid Saad Al-Harbi Treatment-resistant depression: therapeutic trends, challenges, and future directions Patient Prefer Adherence. 2012; 6: 369–388.