loader
About Hyperfine

Software Analytics for Precision Medicine

Precision medicine is where companies use data to better treat, diagnose, and prevent complex diseases. Hyperfine's software is designed to help life science companies transform data into life-changing insights. Our software methods address three different categories of analytics: network analytics, literature analysis, and interdependent variable prospective research. We deliver innovative technologies to help institutions more quickly develop safer and better therapies for patients. Hyperfine's software can help life science institutions utilize the potential of big data to discover, develop, and commercialize transformative, new solutions from:

  • Multi-scale and Systems Biology
  • Precision Oncology
  • Genomics and Multi-Omics Data
  • Imaging Data
How it works

How Our Platforms Work

fancybox
Multi-Layer Analysis

Multi-Layer Analysis helps researchers develop novel populations for specific research. The software creates these by a number of functions at each layer including generating criteria from one population, at one layer, and used as criteria to gather from a second population, on the next layer, or generated by evaluating variables and changing the population at each layer based on the evaluation from a rolling set of variables. This enables the discovery, for specific use by a research team, of novel-unique populations of interdependent variables.

fancybox
Comprehension Normalization Method for Networks

The Comprehension Normalization for Networks uses network data to rapidly extract subnetworks. Subnetworks found from hundreds of network scenarios can elucidate the value of each other. Rapid extraction of subnetworks, and using them with each other promotes a sort of scaling up of systems biology to address a second order of information discoverable with the many subnetworks in context. CNM for networks acts over 1000 times faster than integrative methods at extracting functionally related subnetworks.

fancybox
Comprehension Normalization Method for Language

The Comprehension Normalization for Language interprets the literature from a particular field from the nuances of another field. The result breaks down silos between fields and germinates cross-disciplinary findings, quickly and easily. CNM for language recognizes that language is a network, and the software uses differences in the word associations from two fields to be able to reconfigure the content (from the first field’s language) from the perspective of the second field.

drive02
Validations and Pilots

Comprehension Normalization Method for Networks

CNM for Networks in Obesity

Tissue-To-Tissue Coexpression Genomic Data

Our first validation of the Comprehension Normalization Method for Networks was on gene coexpression genomic data – specifically cross tissue (Tissue-to-Tissue) coexpression between the Hypothalamus, Liver, and Adipose tissues in obese mice.

Discovering the pivotal subnetwork that is most correlated to changes in the mouse obesity traits from the coexpression clusters took the original scientists 2 months. Our software was able to accomplish that same subnetwork in 4 minutes.

 

  • Fewer Manhours: 1 person for 4 minutes, vs. 2 people for 2 months
  • Much Less Computing Power: 1 laptop for 4 minutes vs. Mount Sinai's Minerva, a supercomputer for 2 months
  • Less statistical uncertainty

CNM for Networks in Schizophrenia

Schizophrenic Patients and Controls doing an N-Back task, or being at rest.

With no additional information than clusters made from parts of the brain active together, CNM was able to de novo identify and extract the visual subnetwork, and the default mode subnetwork.

  • Needed no other information about the brain than these two sets of clusters of the brain areas active together (one set made while the patients were at rest and the second made while the patients were performing an N-Back task)
  • CNM works on imaging data: the networks and clusters were generated from fMRI images of the brains

CNM for Networks in Alzheimer's

Using the 19 networks from 19 brain regions, we were able to discover novel communication patterns between the different areas of the Alzheimer’s brain. We accomplished this analysis (19 regions with each other – 342 analyses) in 2 weeks; the original integrative methods would have taken 60 years.

Late Onset Alzheimer's Disorder (LOAD) Brains from the Mount Sinai Brain Bank

  • Done in weeks vs. decades - a new level of relationship research is possible
  • Using the regions but all with each other we were able to discover cross region communication - which helped to characterize the regions too
  • Radically expanded capacity to study the brain
drive02
Awards

Awards & Accreditations

drive02
fancybox
ELabNYC
fancybox
First Growth Venture Capital (VentureCrush)
fancybox
SBIR Impact
Blog

Recent Blog