Agentic education · Learning science

Actionable companion · Motivation & engagement

Learning with agency: Evaluation of KU-F40 automated microscope for parasitology: when artificial intelligence meets old school microscopy

AI-assisted editorial implementation guide, separate from the original research. Proposed actions are not claims or recommendations attributed to the source authors. No independent peer review of this guide is asserted.

Read original paperSearch paper agentCC BY 4.0

Evidence and context

Title and abstract screening; stage labels describe reported settings, not verified numerical ages. Design labels require full-text appraisal. Reported stage(s): Age / stage not established. Verify the participant population and task before transferring the findings.

Original authors’ abstract, reproduced under the linked source license:

ABSTRACT Intestinal parasitic infections (IPIs) have a worldwide distribution and have a major impact on health, work capacity, and economy in many countries. Light microscopy is still considered the reference method for IPI diagnosis but is labor-intensive. KU-F40, an automated feces analyzer, combines automated microscopic examination of stool samples and deep learning artificial intelligence. The aim of this study is to evaluate the performance of KU-F40 for the diagnosis of IPI. A random collection of stool samples prescribed for IPI investigation was retrospectively collected from six clinical laboratories in Belgium along with external quality controls. All samples were analyzed in our laboratory by wet mount preparation using classic light microscopy as reference. We assessed the sensitivity and specificity for parasite detection/identification. Finally, we studied the improvement in parasite detection rate when increasing the number of pictures taken to 150% and 200% of the standard settings. A total of 267 clinical stool samples were included. Using standard settings, overall sensitivity and specificity were 86% and 45%, respectively. When considering only clinically relevant parasites, sensitivity was 95%. Increasing the number of pictures allowed to improve detection rate, but it remained under 90% for several targets. KU-F40 offers an innovative approach and provides welcome automation in the diagnosis of IPI. Currently, its performance does not allow it to be used as a screening tool with automatic validation of negative results. Critical missing features could enhance its performance, including the addition of a 10x magnification objective and additional parasites currently absent from the database. IMPORTANCE Intestinal parasitic infections have a worldwide distribution and are a global health concern in many countries. Light microscopy is still considered the reference method for diagnosis but is labor-intensive, time-consuming, and requires highly skilled and motivated technologists. In this paper, we evaluate the KU-F40, an automated feces analyzer designed to diagnose intestinal parasitic infections by combining automated light microscopy and deep learning artificial intelligence for detection and presumptive identification of several protozoans and helminths. As it relies on microscopy, this method enables the detection and identification of a predefined panel of parasites, whose morphology is known to the system and included in the database, without requiring prior diagnostic suspicion, similarly to multiplex PCR assays. The automation could improve the quality, standardization, and turnaround time of stool parasitology. This study is the first to evaluate the performance of the KU-F40 on a wide range of parasites, collected from six Belgian hospitals, including our two national reference centers.

What kind of evidence is this?

Supporting AI learning science · Evidence synthesis

Stage: Age / stage not established. Screening labels describe reported context and study-design language. They are not a risk-of-bias rating or proof of effectiveness for an age band.

Study methods · source passage

This study was conducted at the clinical laboratory of EpiCURA Hospital Center, an 806-bed secondary-care facility serving the Mons-Borinage and Ath regions of Belgium.

Limitations and uncertainty · source passage

This study has several limitations. First, our sample set is not representative. The number of certain parasites tested was low, as they are rarely encountered in our country, which limits the ability of our study to fully assess the performance for each parasite tested. Additionally, a key and frequently observed parasite, Cryptosporidium spp., was absent. Although we collected Cryptosporidium-positive samples, this parasite was primarily detected by rapid antigen testing rather than manual microscopy. Therefore, we chose not to include these samples in the present study. However, it would be interesting to validate the KU-F40 Cryptosporidium antigen test, not available at the time of this study. Sensitivity could be significantly improved by testing together the microscopic module alongside antigen testing for Giardia and Cryptosporidium, two of the most commonly encountered parasites in our country. Another limitation is that only positive samples were reanalyzed with an increased number of images. This allows us to assess the impact on detection rates but not on specificity. The limitation concerning our contamination study has already been discussed above. We underline that a larger number of replicates, including highly loaded samples with different parasites, would be necessary to assess this risk. Finally, in this study, we used the default threshold values established by the manufacturer for each parasite (see the supplementary material at https://doi.org/10.5281/zenodo.18268350). Although not recommended by the manufacturer, these values can be manually adjusted to affect the sensitivity and specificity of the system on a parasite-by-parasite basis. This option was not explored in the present study.

Design the system before the pilot

Use this study to inform a learning mechanism or implementation question. It is not direct evidence that autonomous agents improve learning. Keep the proposed agent behavior separate from what the source actually tested.

  • Measure engagement separately from achievement. Check whether novelty-driven participation persists and whether it translates to independent performance.
  • Use common rubrics, blind scoring where feasible and a fresh unaided reasoning task. Document permissible assistance and avoid treating AI-detector scores as proof of misconduct.

A 90-day district pilot

  1. Days 1–15 · Appraise. Have the curriculum lead and a teacher read the methods and limitations. Record the study design, sample, comparator, outcomes and evidence gaps. Define a local question that the study can help inform.
  2. Days 16–30 · Design. Treat the paper as background evidence until its population is verified in the full text. Do not assign an age-specific intervention or extend adult results to children. Name an accountable owner and set the baseline, success threshold and budget before implementation.
  3. Days 31–60 · Test. Start with a small voluntary cohort, preserve an accessible non-AI route and require teacher review. Use synthetic or de-identified examples; approve data handling and tool terms through district processes.
  4. Days 61–90 · Decide. Extract the participant ages or education stage, recruitment, task and comparator before selecting a local pilot. If these cannot be established, retain the paper for design discussion rather than age-band efficacy claims. Compare with baseline, document uncertainty and stop or revise if learning worsens, access is unequal or critical errors persist.

The board decision brief

Report the learning problem, evidence quality, participating population, total cost including review time, measured benefit, errors, access differences and remaining uncertainties. Make a dated recommendation to stop, revise or expand. A positive perception survey alone does not establish learning improvement.

Attribution and reuse

Antoine Aupaix; Lorenzo Filippin; Justine Jaumot; Stéphanie Cannoot; Monia Chemais; Delphine Martiny; Véronique Yvette Miendje Deyi; Marine Deffontaine; Corentin Deckers; Valérie Verbelen; Idzi Potters; Charlotte Drieghe; Samy Mzougui; Reza Soleimani; Patrick Philippart; Jonathan Brauner. Evaluation of KU-F40 automated microscope for parasitology: when artificial intelligence meets old school microscopy. Journal of Clinical Microbiology (2026-02-10). https://doi.org/10.1128/jcm.00743-25

Companion prepared 2026-10-30. Original authorship and DOI are preserved. The abstract is an attributed source extract; the implementation plan is new AI-assisted editorial material, offered under the same CC BY 4.0 license. This guide has not undergone independent peer review.