The Alto Advantage: Let Data Do the Talking
The Alto Advantage is a five-part series that covers the company’s guiding principles for pioneering precision psychiatry and, ultimately, establishing a new treatment paradigm in mental health.
Removing guesswork from psychiatry requires data-driven approaches, and the road to eliminating trial-and-error begins with science.
Dr. Wei Wu, Co-Founder and Chief Data Science Officer, has built Alto’s AI infrastructure and drives its algorithm development to provide breakthrough insights into relevant brain biomarkers. For Dr. Wu, the strength of Alto’s platform and data set are the company’s greatest differentiators.
Unlike most biotech companies, Alto is composed of equal parts data scientists and clinical members, and most team members engage with the data in some capacity.
“Thanks to our high-performing clinical team, we collect a large amount of top-quality human biomarker data across multiple collection modalities in our studies,” said Dr. Wu. “The data science team never has a dull moment as we rigorously analyze this data and refine our platform.”
In addition to evaluating incoming clinical trial data, the data science team identifies relevant biomarkers to predict treatment response for Alto’s drugs, as well as supports the technical team to test out various technologies and devices used in Alto’s clinical studies.
He continued, “Almost daily, we’re solidifying and polishing our analysis pipeline in light of new findings and data as they come in. This helps progressively build our confidence in the biomarkers we’ve discovered.”
Dr. Wu notes that while the company is most excited about the end result of helping patients and the potential to stratify patients based on these biomarkers, a conservative approach to proper data capture and analysis is necessary.
Cutting Through the Noise
“With a multitude of AI techniques and numerous ways of analyzing any given dataset, it can be easy to find something in any given dataset,” commented Dr. Wu.
He continued, “It's another thing entirely to demonstrate that the findings are not a fluke or due to overfitting of the data. This is why we never stop validating our predictions – we faithfully listen to what the data say and remain skeptical of findings until we see consistent replications across multiple datasets.”
The Alto team constantly works to improve the validity and reliability of their data features as well as re-test the company’s predictions.
When asked about the most important aspect of biomarker discovery, Dr. Wu eagerly shared, “Data analytic strategies should be deeply tailored for each data type or dataset to generate the most interpretable biomarkers. By designing methods best suited for the data, our biomarkers truly reflect underlying constructs of the disease in an accurate and reliable manner. This requires us to use – or develop, if needed – appropriate methods to collect and efficiently analyze the data.”
Dr. Wu has spent nearly 20 years conducting research in this area as a professor and biomedical engineer with deep expertise in EEG signal processing and novel AI methods.
When partnering with Dr. Etkin at Stanford before co-founding Alto, he shared, “I was interested in moving from academia to industry because I'm very curious about finding practical and meaningful use cases of data science.”
Dr. Wu continued, “Modern data analytic methods are great, but their use should always be driven by important scientific or clinical questions that can make a big impact on the real world.”
He found his answer in the work of pioneering precision psychiatry.
“Before Alto and our founding work at Stanford, I was running my own research lab as a professor in China for a few years. By comparison, things are evolving at a much faster pace here at Alto, which means more near-term and real-world impact for patients.”
Real-world Impact
Similarly to Dr. Wu, each Alto team member brings different expertise, academic and professional backgrounds. While this diversity is an asset, it also creates unique challenges.
“I am very proud of our data scientists, who each bring their own experiences working with prior data sets, data analytic and AI techniques, and unique perspectives,” said Dr. Wu.
“This is extremely helpful as we rethink everything by assimilating useful ideas from other fields, and by testing whether prevailing data analytic assumptions are optimal for psychiatry. The only way to find out is to empirically test – and retest – these ideas on our data.”
While academia is typically defined by a straightforward goal of publishing research papers in high-impact journals and receiving grant funding, the Alto team understands creating real-world impact through clinical studies, and the eventual approval of novel therapies, is another thing entirely.
“In order to obtain results for papers, one often analyzes the same dataset in many different ways and tends to only report positive results in the paper while ignoring negative ones, which has contributed to the so-called replication crisis. Aside from being unethical, this research model simply doesn’t work in biotech,” remarked Dr. Wu.
He shared, “At Alto, our performance metric is clear and simple – successful use of our biomarkers to enhance treatment response in randomized clinical trials. This approach will show our biomarkers in action and, since all analyses are pre-specified, there is no room for skewing results.”
Alto boasts having the largest and most advanced human datasets in precision psychiatry, but Dr. Wu notes, “Compared to other fields that may have millions of samples in a single set, like computer vision, we’re working with relatively small data sets.”
What does this mean? “We are much more careful about how we analyze our data and do not place our fate entirely in the hands of AI; rather, our data analytics are guided by deep domain knowledge and extensive prior research encompassing basic science and translational studies,” said Dr. Wu.
Dr. Wu concludes, “Validating that the biomarkers we’ve identified consistently work for our drugs will fundamentally change the way patients are treated in the real world. That’s the ultimate goal we're trying to achieve.”
As Alto continues to let data do the talking, the company will remain guided by facts and grounded in science.
----------------------------------------------------
If you are a doctor or clinic interested in learning more about participation in our drug trials, please send us an email at trials@altoneuroscience.com.