Identification of Type 2 Diabetes Subgroups through Topological Data Analysis of Patient Similarity

Ayasdi’s collaborators delivered another significant win yesterday, with Mt. Sinai’s Ichan School of Medicine getting published in the ultra-prestigious journal Science: Translational Medicine for their work using topological data analysis to identify previously unknown diabetes subtypes.   

The paper contains breakthrough research and is a validation that precision medicine translates (no pun intended) beyond cancer.  Dr. Joel Dudley, Director of Biomedical Informatics at the Icahn School of Medicine at Mount Sinai used Ayasdi Core to identify new patient subgroups, which will ultimately enable more precise diagnosis and therapies for this widespread, expensive and devastating disease. 

Amazing for science, medicine and patients alike. Diabetes is a prevalent disease with a large incidence, globally and many of us have family members affected by it. 

Physicians know that patients who carry the T2D diagnosis have a variety of phenotypes and susceptibilities to diabetes-related complications.

The Icahn School of Medicine at Mount Sinai has a large database that pairs the genetic, clinical, and medical record data of over 30,000 patients. Mount Sinai’s data set is not only massive, but also complex since it contains many different data types. In addition to genomic sequencing data, it also includes information about each patient’s age, gender, height, weight, race, allergies, blood tests, diagnoses, and family history. Joel and his team used a precision medicine approach to characterize the complexity of T2D patient populations based on high-dimensional electronic medical records (EMRs) and genotype data from 11,210 individuals (genetic markers and clinical data, such as blood levels and symptoms).

Using Topological Data Analysis, Mt. Sinai identified three distinct subgroups of T2D in patient-patient networks.

  • Subtype 1 was characterized by T2D complications diabetic nephropathy and diabetic retinopathy;
  • Subtype 2 was enriched for cancer malignancy and cardiovascular diseases;
  • Subtype 3 was associated most strongly with cardiovascular diseases, neurological diseases, allergies, and HIV infections.

SCI 3 Types

Their network graced the cover of the magazine and can be found on the STM website.  

Hence, using Ayasdi Core, they were able to uncover hidden patterns in large and complex datasets to enable innovative research institutions and pharmaceutical companies like Mount Sinai to expedite biomarker discovery, segment disease types and target drug discovery.

Because Ayasdi’s approach does not require the development of extensive clinical hypotheses, and can automatically map relevant patient subgroups based on hundreds of advanced mathematical algorithms, Dr. Dudley’s initial findings demonstrate that type 2 diabetes is not a singular disease.

Instead, it is comprised of several sub-groups, each with their own distinct sets of complicating factors.

This new insight could potentially lead to more effective treatment protocols and better patient outcomes for all type 2 diabetes patients, and advance the practice of precision medicine.

“By using Ayasdi we see that the current clinical definition of type 2 diabetes is too imprecise,” said Dr. Dudley. “Our data indicates that there may be a type 3, 4, 5, or more diabetes based on the sub-grouping of patient populations who could each benefit from different therapeutic approaches. These are exciting findings with potential for transforming how we approach the treatment of this major disease.”

This is a third example of physicians at major hospitals using our software to improve patient care and publishing in top tier journals.

In summary, here are the other two successful collaborators:

In addition to the work on diabetes, the team at Mount Sinai team used Ayasdi to study other major diseases. In schizophrenia research, for example, investigators integrated MRI data collected across hundreds of cases and paired it with molecular pathway data. They generated data models shaped like the brain that reveals a small region that is significantly different in schizophrenia cases when compared to controls. In the future, this finding can potentially allow doctors to identify and treat schizophrenia more effectively by pinpointing which molecular pathways are disrupted in the brains of schizophrenic patients.

Stay tuned for even more news in the coming quarter as our collaborators continue to change the world using our software.