Early diagnosis of disease can quite literally be the difference between life and death. However, current methods of diagnosis are problematic because:

  1. They're invasive (e.g. biopsies, PAP smears etc.) and therefore can put the health of patients at risk,

  2. They're expensive (e.g. the cost of lab work and expert analysis), and

  3. You have to know what you're looking for. If symptoms are non-specific and suggest more than one disease, multiple methods of diagnosis (e.g. CT scans followed by colonoscopies) must be tried, increasing the aforementioned risk and cost. Worse still, some diseases are symptomless until it's too late. Take pancreatic cancer for example, where a tumor grows rapidly without causing symptoms and is often diagnosed when it’s no longer curable. 


Medicine goes molecular

All these issues mean there's long been demand for better, faster, earlier diagnoses. While the existing methods are still preeminent, advances in molecular medicine are bringing new possibilities to the fore.

Evidence shows that when a malignancy is present, changes occur in the protein composition of biofluids such as blood, urine or saliva. By analyzing these changes, scientists can identify patterns or biomarkers that point to specific conditions. However, to avoid the same pitfalls of traditional methods, any molecular approach needs to be non-invasive,  cheap and work for a wide range of diseases.

One company that is quietly blazing the trail is Miami-based Entopsis. The startup, which was founded by two John Hopkins Ph.D.s, Obdulio Piloto and Ian Cheong, plans to revolutionize how diseases are diagnosed and treated.


How does Entopsis work?

Backed by Peter Thiel's Breakout Labs, Entopsis is using nano-engineering and machine-learning to create a single platform that can screen for multiple medical conditions at once, in an apparently cheap and convenient way. 

An Entopsis prototype

By contrast, says Piloto, current technology requires use of a mass spectrometer,  a freezer-sized machine that can cost $100,000.

Mass spectrometry

Entopsis’ platform consists of two main parts:

The Entopsis method

1. NUTeC (Nanoscale Unbiased TExtured Capture)

A NUTeC is a piece of glass that has been designed at a nano-scale to capture molecules from a biofluid sample such as blood. NUTeCs, which are small enough to fit in the palm of your hand, feature millions of tiny pockets of different shapes, sizes and textures. These pockets are located within a number of larger circles, visible to the eye, which are called “surfaces”.

When a NUTeC is incubated in a sample (i.e. inserted and left to soak for a few minutes), each pocket collects all the molecules that fit inside it and binds to them. In this way, the NUTeC groups molecules according to their physical composition. The groupings create a unique pattern on the surfaces of the NUTeC device. This pattern is a sort of fingerprint, the sample’s molecular profile or “signature”.

Entopsis then applies heat to the device which causes chemical reactions between the molecules and the surfaces. These reactions result in color changes, forming a readable visualization of the signature.

An image of this visualization is subsequently uploaded to the Entopsis cloud server in the form of a photo or scan.


2. Signature Analysis

Signature Analysis is a machine-learning algorithm that enables Entopsis to analyze the molecular signature on a NUTeC.

The algorithm takes the pattern uploaded to the server and compares its signature to that of others in the database to find similar profiles.  

In order to do this, the algorithm must be taught what a particular condition looks like. This is a matter of analyzing a certain number of samples for that condition, to the point when it can distinguish between those with the condition and those without. Piloto says that number will depend on the condition. In the case of an auto-immune disease for example, the algorithm needs to process 1-200 signatures.


Beyond human diagnosis

Entopsis' platform is a universal approach that requires no customization. Because NUTeCs don't look for any specific biological indicator, the range of molecular data they can collect from samples is wide open. This means that NUTeCs can diagnose all sorts of diseases, in both humans and animals. They can also provide prognoses and assess the probability of relapse.

While the startup’s main focus areas are currently medical - oncology, autoimmune disorders etc. - its broad approach means that the technology has many other applications. This could include detecting salmonella in food, assessing the quality of water or even determining whether wine has aged sufficiently.

It’s just a matter of teaching the Signature Analysis algorithm what it needs to look for.


Mass market and minute clinics

A NUTeC prototype

Having tested out many shapes and sizes, Entopsis is now at the stage of finalizing the design of its NUTeCs. The next step will be to work with different partners to try out new uses of the device. Given its many applications, these partners could be researchers, clinicians, occupational health organizations or even beer-brewers. Piloto says the company is open to approaches and ideas.

Entopsis even has a consumer model in mind, where individuals could use NUTeCs for “minute clinics” to assess their own health or check if tap water is drinkable.

The company will be raising a new round in the near future to facilitate this growth with the eventual goal of making the Entopsis platform a mass-market reality.