Skip to content

Overview

Mişko manages preclinical behavioral studies on laboratory rodents from a single place: the study design, the animals and what they received, the filmed test sessions, the video analysis and the reviewed results.

Biology and neuroscience laboratories that test whether a substance reverses a deficit in a disease model with behavioral paradigms, and need a reliable record of which animal was in which group, what it received, how it was tested, and what the result was, without managing raw video by hand.

  • Substances and disease models: substances (compounds, reference drugs, vehicles, inducing agents) and disease models, shared by every study of the laboratory. Each has its own page in the panel.
  • Paradigm: a scientific test type (Morris water maze, novel object recognition, pole test and treadmill). Defined in code in published versions that never change.
  • Environment: a physical apparatus of one paradigm, with measurement revisions. In a test, an animal is placed in an arena, which is one environment revision.
  • Subject: the animal (code, species, sex) and its body weights.
  • Experiment: the study. It is designed as a draft (groups, regimens, timepoints, protocols, endpoints), started on day 0 and completed.
  • Group: an arm of the study with a role (healthy control, sham, disease control, positive control or treatment), a disease model and a regimen. An animal is allocated to exactly one group, and can be excluded with a reason.
  • Test: one filmed session at one timepoint on one protocol step, with one video and one or more animals, each in its own arena.
  • Result: the analysis of one animal in one test. Every result is accepted or rejected by a person before it counts. Results are data, not a pass or fail verdict.
  • Paradigm trial: a paradigm tried on one video in one arena, outside any experiment, with its own calibration if needed and as many analyses with changed thresholds as wanted. It has no groups, review or report.
  • Report: per endpoint, the group statistics and the comparisons that follow from the group roles (model validity, assay sensitivity, efficacy with rescue %, dose response).

Mişko is deployed on-prem, one laboratory per installation. There is no public sign-up: a superadmin is created by the setup command (the system generates a strong password and prints it once), and all other users are created from inside by an administrator. Access is governed by role-based permissions tied to each person’s role in the lab.

The API does not run computer vision itself. Separate analysis workers claim one animal of a test at a time, track it in the video and upload a trajectory and an annotated video to object storage. The API verifies the uploads and computes every metric from the trajectory, so each measurement has one definition. Video files stay in object storage; the database holds records, metrics, events and references to objects. See Architecture.