Andrew Sage

Assistant Scientist with the Toronto Lung Transplant Program. Published machine-learning, sensor and digital-twin work, mostly on donor lung assessment during ex vivo lung perfusion rather than on CLAD outcomes.

As of 2026-10-10. Primary source: ORCID record.

Summary

Andrew T. Sage is an Assistant Scientist with the Toronto Lung Transplant Program at the Toronto General Hospital Research Institute and an Assistant Professor at the University of Toronto. The UHN Research profile names the use of artificial intelligence and machine learning to integrate multiple biomarkers as one of three programme themes. He is included in this category because the profile states that theme and because opened abstracts describe machine-learning models of which Andrew Sage is an author.

Work summary

Andrew Sage applies machine learning and engineering methods to the assessment of donor lungs. Ex vivo lung perfusion (EVLP) is a technique in which a donor lung is kept working outside the body so that it can be assessed before transplant. A machine-learning model called InsighTx, trained on 725 EVLP cases, predicted outcomes after transplant, and in a blinded retrospective study it was associated with a higher likelihood of transplanting suitable lungs and a lower likelihood of transplanting unsuitable lungs [DOI 10.1038/s41467-023-40468-7]. A convolutional neural network, a type of image-analysis model, was trained on 1,300 radiographs from 650 EVLP cases and modestly improved the classification of clinical outcomes [DOI 10.1038/s41746-024-01260-z]. Andrew Sage is also the last author of a 2025 review which notes that several groups have built models for outcomes including chronic lung allograft dysfunction (CLAD) [DOI 10.1097/MCP.0000000000001168]. Separate engineering work describes a microfluidic sensor platform for continuous glucose and lactate monitoring during EVLP [DOI 10.1002/advs.77123] and a digital-twin framework, a computational model of the lung, built from a large clinical EVLP dataset [DOI 10.1038/s41587-026-03121-4].

Based on 3 or more opened abstracts. Written only from abstracts that were opened (PubMed or Europe PMC); full texts were not read. Plain-language explanations are added by the editors and are not from the papers.

Role and field

Role: Assistant Scientist; Assistant Professor (University of Toronto)
Role status: verified from official institutional profile (UHN Research)
Role basis: UHN Research profile, opened on 2026-10-10: Assistant Scientist with the Toronto Lung Transplant Program; Assistant Professor, Department of Surgery, University of Toronto.
Field: machine learning for donor lung assessment; biomarker and assay science
Field basis: Programme themes stated on the UHN Research profile and the abstracts of the source papers.

Only what a cited source supports is stated. Entries marked self-asserted come from the person's own ORCID record. The field is taken from the name of the unit, not inferred from the person's name.

Details

affiliation: Toronto Lung Transplant Program, Toronto General Hospital Research Institute, University Health Network; University of Toronto
ai evidence: UHN Research profile, research interest 3: use of artificial intelligence and machine learning to integrate multiple biomarkers and augment clinical decision-making. Opened abstracts of the five source papers.

Sources and links

Links from this record

Known gaps in this record

Most of the machine-learning work concerns donor lung assessment during ex vivo lung perfusion; the abstracts read do not report CLAD outcomes.
The sensor paper (MetaboSense) describes an assay and microfluidic platform, not a machine-learning method.
The digital-twin paper is a computational modelling study of ex vivo lungs, not a CLAD study.
Current funding was not recorded.
Full texts were not read.
Moved from the People category and not part of the ranked researchers.

Information resource only. Not medical advice. Not a substitute for the care of the patient's transplant team.