A team of data scientists and physicians worked in collaboration in order to provide an AI tool that serves medical professionals with their task of B-NHL immunophenotyping. Different B-NHL entities need different therapies, but the correct diagnosis may be challenging in some cases. The access platform was termed PLAIT.

It is our vision that our XAI fills the gap of consultation and teaching by highly experienced immunophenotyping specialist because it was designed open and transparent (Ref Lit). Flow XAI claims its certainty in three levels and is capable to explain and visualize its decision. Correct decisions serve physicians and patients.

Specialist may submit data files in either fcs or lmd format. In order to avoid transfer of sensitive patient data we provide an anonymization tool (AnonApp) that can be applied before data upload. Therefore PLAIT is inline with the requirements of the general data protection regulation (GDPR).

At the moment there is no international consensus B-NHL immunophenotyping panel, but the underlying FlowXAI is capable to learn even from limited training sets (Ref Lit). Particularly small and intermediate diagnostic centres may therefore benefit from the service of PLAIT.

FlowXAI has already been trained with two independent data sets (Ref Lit) and we provide sample data to become familiar with PLAIT functioning.

A second training cohort from the medical school centres Erlangen and Marburg (Germany) will soon be available; the underlying B cell panel (2 tubes) has been optimized as described previously (Ref. Lit Hoffmann + Thrun et al.) but has no IVD certificate yet (hier Link zu unserem neuen B-NHL Panel Design.

PLAIT is intended to be open source, you are welcome to submit comments that may improve the XAIs knowledge data base (KDB). Please contact us via e-Mail

Please be aware, that PLAIT is NOT a medical device.

As the KDB increments the performance of FlowXAI can be optimized and may increase. Any suggestions or scientific input is appreciated.