0.7.2 · local first
Scholion connects laboratory history, recorded treatment, genetics and daily life. See the long-term direction, notice a recent reversal and bring a focused summary to your clinician.
Install Scholion 0.7.2 to use system insights, the annual overview and the focused visit brief. Your existing records stay local.
0.7.2
Explore gene function and variant evidence within the existing body Systems. Compare clinical estimates with their dated inputs, and inspect sensor observations alongside laboratory history.
525 genes and a separately labelled gene region. Methylation has shared navigation; research associations and clinical neurogenetics remain separate. GenCC validity and variant confidence keep their own labels. All 1469 main-catalogue genes retain sourced function descriptions, including those outside these themes.
FIB-4 and PhenoAge share a compact Today block. Each method keeps its outcome, applicability and sources; missing measurements are named and are not borrowed from another draw.
Preview a Lumiflex application PDF before importing. Printed statistics, logged events and approximate curve readings remain distinct. Revisions remain accessible, and sensor glucose never replaces laboratory glucose or HbA1c.
Choose systems, gene panels and the question for the appointment. The short handout follows body systems needing attention and brings together dated values, available annual trends and relevant prescriptions. Each genetic factor states its link to the discussion. Full details stay in the product.
Findings show the actual reading, zygosity and sourced inheritance. A supported recessive carrier finding stays distinct from disease. Gene biology comes first; ClinVar classification, GenCC validity and Scholion confidence keep separate meanings.
Repeated lookups are shared within a reading and startup document checks run separately from the web interpreter. Existing measurements and genetic reading safeguards remain in place; this change needs no recalculation.
0.7.0
A lower value is not always an improvement. Scholion checks the direction against recorded targets or laboratory ranges, keeps the year separate from the last quarter and month, and shows which observations support the finding.
The person map keeps body systems in view. Compact insight cards distinguish a long-term trend from emerging changes and show where the direction has recently reversed. Detailed readings remain one click away.
Visit mode opens an organised sheet: findings by system, supporting charts, specific discussion questions and the recorded current regimen. Finished courses stay in history. Sources and reference genetics are available when needed.
Record a dated change in symptoms, function or tolerability. Review it alongside measurements before and during a recorded treatment course. A reported response and a measured change remain different kinds of evidence; neither alone proves treatment caused it.
Where a supported, read variant association matches the measured phenotype, Scholion adds the genetic evidence and its confidence level. Known lifestyle measurements provide context. A predisposition does not turn a normal laboratory result into an abnormal one. Clinicians can expand source links, call quality, applicability and observations in the other direction. Hypotheses remain explicitly labelled. Clinicians can compare several measurements with a diplotype and explore mechanisms across broad panels, including amino-acid, B12, lipid and phosphate metabolism. Reference hypotheses remain available when personal evidence is incomplete; they do not change the system score or establish a diagnosis.
Recorded weight and wearable weight share a source-labelled timeline. BMI follows those same weights and the recorded height. Device monthly means remain distinguishable from individual readings, including in charts and assistant responses.
The web app, CLI and MCP use the same findings. An assistant can retrieve the overview, build the visit agenda and compare with an explicitly saved review. Changed observations and revised interpretations remain distinguishable.
Personal mode helps you follow changes between visits. Visit mode brings the same facts into an agenda for the conversation with your clinician.
These videos show the previous interface. The screenshots above show the 0.7.0 interface. The videos are separate from release packages.
The local profile stays on your computer. If you share Scholion results or collected context with a hosted assistant, that information goes to its provider.
Start with a synthetic demo, then connect your own sources. A genome is optional: laboratory history and prescriptions are useful on their own. The held reference base links 1469 genes across sixteen body systems; this is a reference catalogue, not a count of diagnoses.
Read PDF forms or enter a result by hand. Preserve the measurement date, unit and the reference printed on the form; a missing reference stays missing.
Read VCF and consumer array exports, including supported archive wrappers. With local tools, a BAM can show mitochondrial variants, their fractions, reading quality and an estimated maternal lineage. A chip does not become a whole genome.
Compare the current regimen with held pharmacogenomic support and measured phenotype. Track sleep, activity and recorded experiments without turning associations into treatment effects.
The local web app and CLI, a model skill, MCP and agent plugins use the same engine. The tool interface registers 41 tools. A model is optional: code computes the result; the assistant can phrase it.
No automatic PGS or whole-genome recalculation is needed for the new analytical views. Stored results remain available. If a newly curated ORAI1 or FOXO3 position is unread, genome updates identifies the missing position; re-genotyping is an explicit operation that needs suitable local inputs.
A single result within its recorded range can be reassuring about that measurement even when there is no trend. Missing trend points alone do not create a request for more tests.
A research and educational tool, not a medical device, diagnosis or prescription. Clinical decisions belong to the patient and treating clinician.
Genotypes do not establish symptoms or treatment. Unread positions, incomplete coverage and missing support are explicit gaps.
Polygenic percentiles are model-dependent positions in a reference population, not probabilities of disease. Many source cohorts are predominantly European.
Core computations run locally. Optional reference lookups are explicit; SCHOLION_OFFLINE=1 blocks network requests. Connecting an assistant is a separate choice about what context it receives.
Python 3.10–3.13 on Linux, macOS or Windows. The PyPI package includes PDF reading; genome preparation uses separately installed bioinformatics tools.
pip install scholion
scholion init --demo
scholion overview
scholion serveUse the demo first. For your own profile, follow the loading guide; these commands do not represent a clinical assessment.
This page describes Scholion 0.7.2. The code is Apache-2.0; the curated knowledge base is CC BY 4.0.