Integrated outbreak intelligence on one map

Eight kinds of data, each in its own shape: boundaries, population, climate, journeys, case counts, clinical records, wastewater readings and genomes. They stream into one map as layers, the layers close into one picture, and one place lights up with the question: what is happening here, and why?

Three case files, each laid out differently, arrive as they are. An AI helper reads their columns, matches each one to the same fields in PathGen's data model and checks the values, flagging the rows it could not place. The cases then become a layer on the map.

On a laptop, a raw genome is read by a genomic language model and becomes an embedding, a short list of numbers, in seconds. Only the embedding is compared with reference embeddings; the raw genome stays on the laptop. Embed another genome and the read starts again, and its embedding finds its own closest match.

Sample embeddings gather into clusters by similarity. Each cluster goes onto the map where its samples were collected, and the map is read against walking time to care: one emerging cluster sits far from any clinic.

Every sample on one map. Filters for lineage, drug resistance and date narrow them; two spatial cohorts are selected and compared side by side on resistance, recent samples and distance to care, highlighting the priority group for action.

Six kinds of source, each in its own shape: genomes, scientific publications, WHO policies, country policies, geospatial features and your own data. They join one graph, and a path runs through it from a sample to the policy that applies where it was found.

A question about samples and places, in plain language. The answer states its limitations first and cites its sources. The evidence map traces each claim to the sources and tools behind it and grades it strong, partial or ungrounded, and the answer ends with recommended next actions.