ENδΈ­ζ–‡

Science

A systems approach to human healthspan

Guided by the bio­logy of NAD+ meta­bol­ism, mito­chon­drial func­tion, pro­teo­stasis and meta­bolic and renal aging, we train advanced cross-spe­cies real-world models on the accu­mu­lated phen­o­typic and multi-modal data of a biobank that we hold under licence.

01

The problem

Aging is the common denominator

Aging is the single largest risk factor for the dis­eases that dom­in­ate medi­cine in later life. Each is stud­ied and treated sep­ar­ately; the pro­cess that makes a body sus­cept­ible to all of them at once is stud­ied far less.

That pro­cess has a shape. Physiolo­gical reserve does not fall in a straight line; it decays. That is why an inter­ven­tion applied early enough, on the right mech­an­ism, does not merely add time: it changes the expo­nent. Our work is dir­ec­ted at that expo­nent.

Healthspan, illustratedCurves of physiological reserve plotted against age. The dashed curve is the steep exponential decline of untreated aging, and four fainter dotted curves fan around it: individual trajectories, since nobody declines at the average rate. The solid curve declines more slowly under intervention, above all of them, and the shaded area between it and the untreated average is the healthspan gained. A network of connected nodes above the curves stands for the biological data and computation used to find such interventions.

02

Method

A systems physiology approach

No single meas­ure­ment explains aging, so we do not rely on one. The plat­form con­nects gen­o­type, molecu­lar state and physiolo­gical out­come across a whole lifespan, and across spe­cies.

  1. 01

    Deep phenotyping

    Genetic ref­er­ence pop­u­la­tions char­ac­ter­ised across the lifespan, from lifespan itself to the meta­bolic and renal phen­o­types that mark age-related decline. Genetic diversity is the point: it is what turns a colony into an exper­i­ment.

  2. 02

    Multi-omic atlases

    Tran­scrip­tomic, pro­teo­mic and meta­bolo­mic layers across the tis­sues where aging is decis­ive, assembled so that they can be read together rather than one at a time.

  3. 03

    Computational modelling

    Sys­tems ana­lysis con­nect­ing gen­o­types to phen­o­types (QTL map­ping and GWAS, PheWAS, medi­ation ana­lysis), with machine learn­ing applied across the whole to sur­face the asso­ci­ations that no single ana­lysis would reach.

  4. 04

    Cross-species validation

    Res­ults are cross-val­id­ated against human cohorts. This seam­less integ­ra­tion of mouse and human data is what makes a find­ing a can­did­ate for clin­ical trans­la­tion rather than a spe­cies-spe­cific arte­fact.

  5. 05

    Target validation and translation

    Can­did­ates enter func­tional val­id­a­tion and a trans­la­tional pipeline groun­ded in tissue-level bio­logy.

03

Focus

The biology we act on

Four mech­an­isms, chosen because they are meas­ur­able, modi­fi­able, and implic­ated across more than one age-related dis­ease.

01

NAD+ metabolism

Couples nutri­ent state to repair, and its avail­ab­il­ity falls meas­ur­ably with age across tis­sues. It sits upstream of a great deal of aging physiology, which makes it both a target and a readout.

02

Mitochondrial function

Bioen­er­getic capa­city and the qual­ity con­trol that defends it. Tis­sues fail in ways that track their ener­getic demand, which is why this recurs across so many age-related con­di­tions.

03

Proteostasis

Pro­tein fold­ing, turnover and clear­ance. Its fail­ure is one of the most con­sist­ent molecu­lar sig­na­tures of aged tissue, and one of the clearest points of inter­ven­tion.

04

Metabolic and renal aging

Phen­o­types from meta­bolic dis­ease through kidney decline, tracked lon­git­ud­in­ally so that tra­ject­ory, not just end­point, is avail­able to the ana­lysis.

04

Intelligence

Where artificial intelligence enters

Bio­logy sets the ques­tions: which tissue, which times­cale, which phen­o­type actu­ally mat­ters. Com­pu­ta­tion answers them at a depth and speed that no labor­at­ory could reach unaided.

A model with noth­ing to learn from pro­duces con­fid­ent noise; a data­set with no one to read it stays a data­set. Our pos­i­tion is that the two have to be built together, which is the logic of our part­ner­ship with Natural Constant.

Cornaro.ai x Natural Constant

The biology that sets the questions, the mathematics that helps answer them.

Collaborate

We work with academic groups, clinical cohorts and industry partners who bring data, models or questions to the same problem.

Start a conversationAbout Cornaro.ai