Joerg Hoffmann

In machine learning, the Bayes classifier represents the theoretical optimum for minimizing classification errors. Since estimating high-dimensional probability densities is impractical, simplified approximations such as naïve Bayes and k-nearest neighbor are widely used as baseline classifiers. Despite their simplicity, these methods require design choices—such as the distance measures in kNN, or the feature independence in naïve Bayes. In particular, naïve Bayes relies on implicit assumptions by using Gaussian mixtures or univariate kernel density estimators. Such design choices, however, often fail to capture heterogeneous distributional structures across features. We propose a flexible naïve Bayes classifier that leverages Pareto Density Estimation (PDE), a parameter-free, non-parametric approach shown to outperform standard kernel methods in exploratory statistics. PDE avoids prior distributional assumptions and supports interpretability through visualization of class-conditional likelihoods. In addition, we address a recently described pitfall of Bayes’ theorem: the misclassification of observations with low evidence. Building on the concept of plausible Bayes, we introduce a safeguard to handle uncertain cases more reliably. While not aiming to surpass state-of-the-art classifiers, our results show that PDE-flexible naïve Bayes with uncertainty handling provides a robust, scalable, and interpretable baseline that can be applied across diverse data scenarios.

In machine learning, the Bayes classifier represents the theoretical optimum for minimizing classifica-tion errors. Since estimating high-dimensional probability densities is impractical, simplified approxima-tions such as naïve Bayes and k-nearest neighbor are widely used as baseline classifiers. Despite their simplicity, these methods require design choices—such as the distance measures in kNN, or the feature independence in naïve Bayes. In particular, naïve Bayes relies on implicit assumptions by using Gaussi-an mixtures or univariate kernel density estimators. Such design choices, however, often fail to capture heterogeneous distributional structures across features. We propose a flexible naïve Bayes classifier that leverages Pareto Density Estimation (PDE), a parame-ter-free, non-parametric approach shown to outperform standard kernel methods in exploratory statis-tics (Thrun et al., 2020). PDE avoids prior distributional assumptions and supports interpretability through visualization of class-conditional likelihoods. In addition, we address a recently described pit-fall of Bayes’ theorem: the misclassification of observations with low evidence. Building on the con-cept of plausible Bayes (Ultsch & Lötsch, 2022), we introduce a safeguard to handle uncertain cases more reliably. While not aiming to surpass state-of-the-art classifiers, our results show that PDE-flexible naïve Bayes with uncertainty handling provides a robust, scalable, and interpretable baseline that can be applied across diverse data scenarios.

Introduction: Inflammatory conditions in patients have various causes and require different treatments. Bacterial infections are treated with antibiotics, while these medications are ineffective against viral infections. Autoimmune diseases and graft-versus-host disease (GVHD) after allogeneic stem cell transplantation, require immunosuppressive therapies such as glucocorticoids, which may be contraindicated in other inflammatory states. In this study, we employ a combination of straightforward blood tests to devise an explainable artificial intelligence (XAI) for distinguishing between bacterial infections, viral infections, and autoimmune diseases/graft-versus-host disease.Patients and methods: We analysed peripheral blood from 80 patients with inflammatory conditions and 38 controls. Complete blood count, CRP analysis, and a rapid flow cytometric test for myeloid activation markers CD169, CD64, and HLA-DR were utilized. A two-step XAI distinguished firstly with C5.0 rules pruned by ABC analysis between controls and inflammatory conditions and secondly between the types of inflammatory conditions with a new bivariate decision tree using the Simpson impurity function.Results: Inflammatory conditions were distinguished using an XAI, achieving an overall accuracy of 81.0% (95%CI 72 – 87%). Bacterial infection (N = 30), viral infection (N = 26), and autoimmune diseases/GVHD (N = 24) were differentiated with accuracies of 90.3%, 80.0%, and 79.0%, respectively. The most critical parameter for distinguishing between controls and inflammatory conditions was the expression of CD64 on neutrophils. Monocyte count and expression of CD169 were most crucial for the classification within the inflammatory conditions.Conclusion: Treatment decisions for inflammatory conditions can be effectively guided by XAI rules, straightforward to implement and based on promptly acquired blood parameters.

Typical state-of-the-art flow cytometry data samples typically consist of measures of 10 to 30 features of more than 100,000 cell “events”. Artificial intelligence (AI) systems are able to diagnose such data with almost the same accuracy as human experts. However, such systems face one central challenge: their decisions have far-reaching consequences for the health and lives of people. Therefore, the decisions of AI systems need to be understandable and justifiable by humans. In this work, we present a novel explainable AI (XAI) method called algorithmic population descriptions (ALPODS), which is able to classify (diagnose) cases based on subpopulations in high-dimensional data. ALPODS is able to explain its decisions in a form that is understandable to human experts. For the identified subpopulations, fuzzy reasoning rules expressed in the typical language of domain experts are generated. A visualization method based on these rules allows human experts to understand the reasoning used by the AI system. A comparison with a selection of state-of-the-art XAI systems shows that ALPODS operates efficiently on known benchmark data and on everyday routine case data.

Autoimmune hemolytic anemia (AIHA) can be life-threatening, if hemoglobin (Hb) levels continue to decline after established treatments with glucocorticoids, rituximab, intravenous immunoglobulins, and plasmapheresis. Impaired regulatory T cells (Treg) are proposed to alleviate AIHA development through decreased binding of CTLA-4 to antigen-presenting cells. Abatacept is a fusion protein with a CTLA-4 domain and is approved for use in rheumatoid arthritis. It mimics the immunosuppressive CTLA-4 effect of Treg. Thus, application of abatacept in refractory AIHA might be reasonable. A 54-year-old woman with known AIHA was admitted to our clinic due to therapy-refractory hemoglobin decrease to 4.0 g/dl. Previously, multiple courses of glucocorticoids, rituximab, azathioprine, mycophenolate mofetil, cyclophosphamide, bortezomib, and a splenectomy failed to stop or stabilize hemoglobin levels and hemolysis. A new immunosuppressive therapy with cyclosporine was initiated and erythropoiesis was stimulated with darbepoetin alfa. Again, therapy failed even though we tried to support immunosuppressive therapy by reducing the amount of pathogenic antibody through plasmapheresis. We stopped the treatment with cyclosporine and applied abatacept instead. After seven days hemoglobin stabilized at 4.3 g/dl and no further red blood cells transfusions were necessary. About one month later hemolysis aggravated again and azathioprine was added to the ongoing abatacept treatment. Finally, the combination of abatacept and azathioprine led to a long-lasting increase of the Hb level above 11 g/dl six months later. Abatacept can be applied to overcome therapy refractory autoimmune hemolytic anemia but should be combined with an additional immunosuppressive medication such as azathioprine.

Background: The International Prognostic Index (IPI) is applied to predict the outcome of chronic lymphocytic leukemia (CLL) with five prognostic factors, including genetic analysis. We investigated whether multiparameter flow cytometry (MPFC) data of CLL samples could predict the outcome by methods of explainable artificial intelligence (XAI). Further, XAI should explain the results based on distinctive cell populations in MPFC dot plots. Methods: We analyzed MPFC data from the peripheral blood of 157 patients with CLL. The ALPODS XAI algorithm was used to identify cell populations that were predictive of inferior outcomes (death, failure of first-line treatment). The diagnostic ability of each XAI population was evaluated with receiver operating characteristic (ROC) curves. Results: ALPODS defined 17 populations with higher ability than the CLL-IPI to classify clinical outcomes (ROC: area under curve (AUC) 0.95 vs. 0.78). The best single classifier was an XAI population consisting of CD4+ T cells (AUC 0.78; 95% CI 0.70–0.86; p < 0.0001). Patients with low CD4+ T cells had an inferior outcome. The addition of the CD4+ T-cell population enhanced the predictive ability of the CLL-IPI (AUC 0.83; 95% CI 0.77–0.90; p < 0.0001). Conclusions: The ALPODS XAI algorithm detected highly predictive cell populations in CLL that may be able to refine conventional prognostic scores such as IPI.

Minimal residual disease (MRD) detection is a strong predictor for survival and relapse in acute myeloid leukemia (AML). MRD can be either determined by molecular assessment strategies or via multiparameter flow cytometry. The degree of bone marrow (BM) dilution with peripheral blood (PB) increases with aspiration volume causing consecutive underestimation of the residual AML blast amount. In order to prevent false‐negative MRD results, we developed Cinderella, a simple automated method for one‐tube simultaneous measurement of hemodilution in BM samples and MRD level. The explainable artificial intelligence (XAI) Cinderella was trained and validated with the digital raw data of a flow cytometric “8‐color” AML‐MRD antibody panel in 126 BM and 23 PB samples from 35 patients. Cinderella predicted PB dilution with high accordance compared to the results of the Holdrinet formula (Pearson's correlation coefficient r = 0.94, R² = 0.89, p < 0.001). Unlike conventional neuronal networks Cinderella calculated the distributions of 12 different cell populations that were assigned to true hematopoietic counterparts as a human in the loop (HIL) approach. Besides characteristic BM cells such as myelocytes and myeloid progenitor cells the XAI identified discriminating populations, which were not specific for BM or PB (e.g., T cell/NK cell subpopulations and CD45 negative cells) and considered their frequency differences. Thus, Cinderella represents a HIL‐XAI algorithm capable to calculate the degree of hemodilution in BM samples with an AML MRD immunophenotype panel. It is explicable, transparent, and paves a simple way to prevent false negative MRD reports.

Hyperinflammation through neutrophil granulocytes contributes to disease severity in COVID-19 pneumonia and promotes acute lung failure. Understanding the mechanisms of the dysregulations within the myeloid cell compartment may help to improve therapies for severe COVID-19 infection. Here, we investigated the immunopathological characteristics of circulating neutrophil granulocytes and monocytes in 16 patients with COVID-19 pneumonia by multiparameter flow cytometry in comparison to 9 patients with pulmonary infiltrates but without COVID-19. We correlated the immunophenotypes with the scores of the severity-of-disease classification system, APACHE-II. We found that the mean fluorescence intensity (MFI) of CD15, which is important for the transendothelial migration, was significantly reduced in the patients with COVID-19 (difference ± SD; 295.70 ± 117.50 MFI; p = 0.02). In addition, the granularity was significantly lower in the neutrophil granulocytes of patients with COVID-19 (difference ± SD; 1.11 ± 0.43 side-scatter ratio; p = 0.02). Moreover, the Fc-gamma receptor III (CD16) and Fc-gamma receptor I (CD64) on the neutrophil granulocytes were expressed discordantly with COVID-19 severity. CD16 correlated as inversely proportional (ρ = (−)0.72; 95% CI (−)0.92–(−)0.23; p = 0.01) and CD64 as proportional (ρ = 0.76; 95% CI 0.31–0.93; p = 0.01) with the APACHE-II scores of the patients. We conclude that the deviant expression of the Fc-gamma receptors might play role in a dysregulated antibody-mediated phagocytosis in severe cases of COVID-19 pneumonia.

Purpose: During neutrophil extracellular traps (NET) formation granulocytes release a decondensed chromatin web that is studded with antimicrobial proteins. These NET engulf and kill pathogens like bacteria and fungi. NET formation is part of the innate immune response but can also contribute to the aggravation of autoimmune diseases, thrombosis, and cancer metastasis. Anti-NET therapeutics to prevent potentially harmful consequences of excessive NET formation are warranted. Materials and methods: Therefore, we stimulated NET formation with ionomycin in the peripheral blood of 25 healthy individuals and quantified NET with flow cytometry and fluorescence microscopy after exposure to five different anti-inflammatory and cytostatic drugs. NET were identified by their expression of myeloperoxidase, citrullinated histone H3, and (extracellular) DNA release. Results: The preliminary in vitro drug screening indicated that acetylsalicylic acid (ASA) might suppress (−3.82%), and rituximab might enhance (+10.52%) NET formation. To consolidate the screening results, we quantified NET after exposure to rituximab and ASA in the blood of nine additional healthy subjects. Rituximab showed a significant increased NET formation compared to the neutrophils treated with ASA (a mean of differences 3.96%; 95% CI 1.90–6.03%; p < .01) or compared to neutrophils without treatment (a mean of differences 4.39%; 95% CI 1.17–7.61%; p = .01). Contrary to the screening results ASA showed no significant suppression of NET formation in the consolidation experiments (a mean of differences 0.43%; 95% CI −1.27 to 2.12%; p = .58). Conclusions: We conclude that rituximab therapy might further trigger activated NET formation and should be applied with caution in patients with pro-inflammatory state and underlying autoimmune disease, thrombosis, or cancer.