Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-1866
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-18T22:02:34.147001Z
status: published
lens: g_space
sector: science
headline: Sustainable hemocompatibility prediction of electrospun nanomaterials using SEM-driven CNN ensemble learning aligned with ISO 10993
dek: Background Electrospun nanocomposite biomaterials are promising candidates for blood-contacting medical devices owing to their high surface area, tunable nanoscale morphology, and biomimetic replication of the native extracellular matrix. However, conventional hemocompatibility assessments rely heavily on biological assays. These methods require blood samples, specialized reagents, and consumables, thereby inherently increasing costs, resource demands, and laboratory waste. To mitigate these limitations, sustain…
gain_title: Fine-tuned CNN models classify nanofibre SEM images as haemocompatible or not, offering a pre-screening step that could cut blood-based assays and lab consumables.
problem_title: (none)
trace_subject: (none)
gain_reading: Fine-tuned CNN models classify nanofibre SEM images as haemocompatible or not, offering a pre-screening step that could cut blood-based assays and lab consumables.
gain_evidence: significantly reduces reliance on blood-based assays and associated laboratory consumables
problem_reading: (none)
problem_evidence: (none)
quick_read: Researchers fine-tuned three pretrained image-recognition networks (VGG19, ResNet50 and InceptionV3) to label scanning electron microscopy images of electrospun nanofibres as haemocompatible or non-haemocompatible. They used augmentation to strengthen the data and a majority-voting ensemble to improve reliability.

The work matters because standard hemocompatibility testing needs blood samples, reagents and consumables, which add cost and waste. An image-based pre-screen could lighten that load for developers of blood-contacting devices. The source frames it only as a pre-screening aid that supports later biological evaluation, not a replacement for it.
limitation: The authors describe the method as a pre-screening tool that supports, rather than replaces, subsequent biological evaluation.
tag: Evidence-backed gain
key_points: InceptionV3 was the best single model, with precision 0.98, recall 1.00 and F1-score 0.99. | The majority-voting ensemble reached an overall accuracy of 99%. | LIME and Integrated Gradients were used to check which image features drove predictions. | The approach is positioned as an eco-friendly pre-screening tool aligned with ISO 10993.
rundown: The study targets electrospun nanocomposite biomaterials, which are considered promising for blood-contacting medical devices because of their high surface area, tunable nanoscale morphology and resemblance to the native extracellular matrix.

Conventional assessment relies heavily on biological assays that need blood, specialised reagents and consumables. The authors argue this raises costs, resource demands and laboratory waste, motivating evaluation based on physicochemical features.

Explainability analysis with LIME and Integrated Gradients indicated that predictions were driven by physiologically relevant microstructural features of the nanofibres. The authors conclude the framework supports efficient, regulatory-compliant biomaterial development by guiding subsequent biological evaluation.
sources:
- peer_reviewed | PLOS One | https://doi.org/10.1371/journal.pone.0359594 | 2026-10-05
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

f597e695a80d1d5b57f9c9f0847287d9cc79710c435c7ed861271de1eb0bda2c

previous

0000000000000000000000000000000000000000000000000000000000000000