CHICAGO – With the American Psychiatric Association's annual meeting set to begin Saturday in New Orleans, Tempus is unveiling a research and precision medicine program to apply real-world molecular and clinical data to neuropsychiatry and other mental health services.
In the program, called Tempus N+, the Chicago-based bioinformatics and molecular laboratory company will work with its network of academic medical centers to collect, curate, structure, and harmonize large amounts of data to build multimodal research pipelines. The firm expects Tempus N+ to break down data silos to facilitate more complete psychiatric research, much as it has done with cancer in the past and is now trying to do in pathology.
The N stands for neuropsychiatry, according to CSO Joel Dudley.
In the new initiative, the firm wants to collaborate with academic medical centers across the US to create a sustainable infrastructure for advanced mental health research. In the company's vision, users will be able to contribute real-world genotypic and phenotypic data in a secure environment to help build a multimodal, longitudinal dataset that is more comprehensive than current options.
"It has always been the mission of Tempus from day one to open up [our] critical mass of data to researchers," Dudley said. "Why now for neuropsych? Basically because we're just starting to get that critical mass of data."
The strategy is a familiar one for the company and its customers.
Tempus entered the pathology market in March by introducing Edge, a platform that provides pathologists with online access to artificial intelligence models to identify samples with potentially significant biomarkers with a single hematoxylin and eosin-stained slide. The platform is meant to help physicians decide whether to order genomic sequencing that might indicate eligibility for targeted therapies and clinical trials.
At the same time Tempus unveiled Edge, the company also announced NAPA, a planned network of pathology laboratories across the US to contribute data to improve the firm's artificial intelligence models, again with the goal of identifying patients for targeted therapies and clinical trials. NAPA is analogous to Tempus Integrated Molecular Evaluation (TIME) Trial, a 3-year-old network that provides similar services for cancer patients.
For N+, Tempus will lean on its experiences with its core oncology platform and will build the dataset from its nP whole-exome pharmacogenomic test, as well as from integrations with electronic health records systems.
"I think it really follows the model that we took in oncology," Dudley said, because N+ also seeks to generate and organize large-scale data from clinical testing. "And then if we have that data at scale, there's a couple [groups of] stakeholders who can benefit," namely clinicians and researchers.
Dudley described nP as a multimodal test because Tempus gathers molecular data through whole-exome sequencing and combines the lab results with phenotypic data. "That's already out in the world and being collected and feeding into this [N+] resource," he said.
There are some important distinctions between the earlier programs and N+, however.
Dudley said that precision neuropsychiatry must rely heavily on patient-reported outcomes data, largely because there are no standard methods of describing disease progression as there are in cancer with its disease stages. "That's why it's really important to enrich the clinical data we get with this other [patient-reported] data that we're collecting, which then in turn should provide phenotypes and data streams to power research in ways that you can't [do] with the clinical data alone," he explained.
Tempus did have to build a new technology stack on top of its core informatics platform to support this data that will feed into N+ as well as neuropsychiatry-specific analysis. "Now that we've done that to scale for ourselves with the hundreds of engineers we have on staff, we really want to open that technology up to neurology or neuropsychiatry researchers because there simply isn't this type of technology available to them, even in the research realm," Dudley said.
The firm is also curating the data through a combination of manual and AI-powered automated processes. "What a tremendous amount of work it is to organize and structure the data, then normalize it across institutions and EHR systems and different types of disease areas," Dudley said.
For that, Tempus built a number of tools for users to query molecular and phenotypic data to identify cohorts of patients to answer specific research questions. "These aren't really tools that we sell. These are just value-added tools on top of the technology stack," according to Dudley.
Even the army of network engineers, data scientists, and computational biologists that Tempus employs is not sufficient to mine all the data the firm is collecting. "We just don't have enough people to ask all the important questions we would be able to ask of the data to solve problems here in [addressing] these diseases, so we need to get other people asking questions on this data," Dudley said.
Given that N+ is brand new, Tempus has not signed any contracts with institutions wishing to participate, though the firm has had ambitions in mental health for several years, particularly with pharmacogenomics-driven depression screening. The new program adds a research element that the PGx screening does not offer on its own.
Dudley said that Tempus will open up the N+ dataset and technology platform to any type of translational research, from drug discovery to frontline treatment. While the firm may eventually welcome some pharmaceutical companies as partners, Dudley said that initial marketing efforts will concentrate on academic researchers.
Dudley said that a challenge in neuropsychiatry is that the current method of defining mental illness — the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM 5) — is "pretty archaic" because it is more static and less granular than what precision medicine promises.
"We haven't had the data in neuropsych to begin really thinking about how … we make a precision medicine ontology or nosology of psychiatric diseases," Dudley said. "Hopefully, this dataset would support that."