Chiara Novelli: Are We Validating Laboratory Algorithms as Rigorously as We Validate Laboratory Tests?
Chiara Novelli, Senior Biologist at Legnano Civil Hospital, shared a post on LinkedIn about a recent article she and her colleagues co-authored, published in International Journal of Laboratory Hematology, adding:
“Recently, our work on the implementation and validation of an automated Lupus Anticoagulant interpretation algorithm has been published in the ‘International Journal of Laboratory Hematology’.
The project was developed across a four-hospital laboratory network and focused not only on LA testing itself, but also on a broader question that many laboratories are increasingly facing:
How should we properly validate rule-based algorithms implemented in middleware and LIS environments?
As laboratory diagnostics become more complex, middleware systems are evolving from simple technical tools into true decision-support platforms.
However, while analytical validation is highly standardized, validation strategies for middleware algorithms often remain heterogeneous and poorly defined.
In our experience, key elements for a robust validation framework include:
- Hierarchical and traceable rule design
- Retrospective and prospective validation
- Discrepant-case analysis
- Management of analytical interferences and exceptions
- Continuous post-implementation monitoring
- Preservation of expert oversight for complex cases
Beyond standardization and turnaround time optimization, one of the most interesting aspects was observing how structured workflows can reduce inter-operator variability and support less experienced professionals without replacing clinical expertise.
I believe this topic goes far beyond coagulation testing and may become increasingly relevant for laboratory governance, accreditation processes, and AI-assisted diagnostics.
I would be very interested to hear how other laboratories are approaching these challenges:
- How do you validate middleware algorithms in your institution?
- Do you use retrospective datasets, prospective parallel testing, or both?
- How do you manage traceability and version control of rules?
- Where do you draw the line between autoverification and mandatory expert review?
- Are existing CLSI/ISTH recommendations sufficient for modern middleware ecosystems?”
Title: Design and Implementation of an Automated Interpretation Algorithm for Lupus Anticoagulant Functional Testing
Authors: Chiara Novelli, Arianna Gatti, Flora Ierna, Elena Riccardi, Irene Cuppari
Stay updated on all scientific advances with Hemostasis Today.
-
May 31, 2026, 05:28Nikolaos Androulakis: Inflammatory Burden and Antiphospholipid Antibodies in Lupus Anticoagulant Testing
-
May 31, 2026, 05:28Alysha Arif: Understanding a Complex Case of Postpartum Seizures
-
May 31, 2026, 02:42Katja van den Hurk: Optimizing Iron Management in Blood Donors
-
May 31, 2026, 02:36The New JPAC Website Has Been Live for Nearly a Week – SHOT
-
May 31, 2026, 02:33Jecko Thachil: Is It Possible to Use Machine Learning for Differentiating Inherited Thrombocytopenia from Other Causes?
-
May 30, 2026, 18:12Khokon Miah Akanda: FDA Approves a First-in-Class Hypertension Drug
-
May 30, 2026, 18:11Alessandro Perrella: It is Not Only about “How High the CRP is”, but “in Which Patient” We are Measuring That Value
-
May 30, 2026, 18:11Jasper van Grunsven: Grateful for Our Rare Disease Team and the Commitment That Drives Our Mission
-
May 30, 2026, 18:10Open Registration for the 2026 Thrombosis Canada Annual Education Conference – Thrombosis Canada