BLINK: Behavioral Latent Modeling of NK Cell Cytotoxicity

Iman Nematollahi; Jose Francisco Villena-Ossa; Alina Moter; Kiana Farhadyar; Gabriel Kalweit; Abhinav Valada; Toni Cathomen; Evelyn Ullrich

cs.CVMar 5, 2026

BLINK: Behavioral Latent Modeling of NK Cell Cytotoxicity

Authors: Iman Nematollahi, Jose Francisco Villena-Ossa, Alina Moter, Kiana Farhadyar, Gabriel Kalweit, Abhinav Valada, Toni Cathomen, Evelyn Ullrich,

Organizations: Department of Computer Science, University of Freiburg, Freiburg, Germany · Institute for Transfusion Medicine and Gene Therapy, University Medical Center Freiburg, Freiburg, Germany · Goethe University, Department of Pediatrics, Experimental Immunology and Cell Therapy, Frankfurt am Main, Germany · IMBIT//BrainLinks-BrainTools, University of Freiburg, Freiburg, Germany ·

Abstract

Machine learning models of cellular interaction dynamics hold promise for understanding cell behavior. Natural killer (NK) cell cytotoxicity is a prominent example of such interaction dynamics and is commonly studied using time-resolved multi-channel fluorescence microscopy. Although tumor cell death events can be annotated at single frames, NK cytotoxic outcome emerges over time from cellular interactions and cannot be reliably inferred from frame-wise classification alone. We introduce BLINK, a trajectory-based recurrent state-space model that serves as a cell world model for NK-tumor interactions. BLINK learns latent interaction dynamics from partially observed NK-tumor interaction sequences and predicts apoptosis increments that accumulate into cytotoxic outcomes. Experiments on long-term time-lapse NK-tumor recordings show improved cytotoxic outcome detection and enable forecasting of future outcomes, together with an interpretable latent representation that organizes NK trajectories into coherent behavioral modes and temporally structured interaction phases. BLINK provides a unified framework for quantitative evaluation and structured modeling of NK cytotoxic behavior at the single-cell level.

Figures & tables

Figure 1: Overview of BLINK : (a) Multi-channel fluorescence microscopy captures NK cells, tumor cells, apoptosis, and morphology. (b) Segmentation and tracking yield NK-centered interaction trajectories. (c) BLINK encodes these sequences into a recurrent latent state-space model that captures interaction dynamics under partial observability, supports latent rollouts for future cytotoxic outcomes, and predicts NK-induced apoptosis increments that accumulate monotonically. The learned latent space organizes NK behaviors into coherent modes.

Model	MAE $\downarrow$	RMSE $\downarrow$	Corr $\uparrow$	Within $\pm 1$ (%) $\uparrow$	F-MAE 30 $\downarrow$
Zero	1.28 $\pm$ 0.16	1.74 $\pm$ 0.15	0 $\pm$ 0.0	54.3% $\pm$ 7.0%	0.12 $\pm$ 0.06
Mean	1.04 $\pm$ 0.07	1.18 $\pm$ 0.08	0 $\pm$ 0.0	49.6% $\pm$ 6.3%	0.24 $\pm$ 0.05
FrameAE	0.95 $\pm$ 0.11	1.14 $\pm$ 0.13	0.32 $\pm$ 0.07	64.9% $\pm$ 6.8%	X
GRU-regress	1.25 $\pm$ 0.14	1.72 $\pm$ 0.14	0 $\pm$ 0.0	55.7% $\pm$ 6.9%	0.12 $\pm$ 0.06
GRU-monotone	0.74 $\pm$ 0.09	1.04 $\pm$ 0.11	0.57 $\pm$ 0.04	71.9% $\pm$ 3.3%	0.22 $\pm$ 0.04
BLINK-no-action	0.80 $\pm$ 0.06	1.14 $\pm$ 0.09	0.61 $\pm$ 0.04	69.4% $\pm$ 7.3%	0.09 $\pm$ 0.01

Table 1: Track-level cumulative cytotoxic outcome prediction on the held-out test set, showing improvements of BLINK across error and forecasting metrics.

Figure 2: Real interaction trajectory (top) and world model-decoded latent trajectory (bottom). Predicted and ground truth apoptosis increments align.

Panel of Figure 3 — Figure 3: Latent behavioral structure of NK trajectories. (a) UMAP of training window embeddings clustered into four modes. (b) Test tracks projected into the embedding. (c) State transition matrix showing temporal mode progression.

Toxicity Detection

Toxicity detection identifies harmful language for moderation and predicts toxic effects of chemicals, drugs, or biological exposures.

See all 46 papers

Understanding Cell Fate Decisions with Temporal Attention
Mar 17, 2026Florian Bürger, Martim Dias Gomes, Adrián E. Granada +2Microscopy Image Analysis Interpretable ML
Institute for Biomedical Informatics, Faculty of Medicine and University Hospital Cologne, University of Cologne, Cologne, Germany · Faculty of Mathematics and Natural Sciences, University of Cologne, Cologne, Germany · Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD), University of Cologne, Cologne, Germany +2
Detection of extremely rare dead cells in fluorescence microscopy images
Oct 9, 2026Prateek Verma, Benjamin Butter, Jian Zhang +1Microscopy Image Analysis Fluorescence Microscopy
Department of Electrical Engineering and Computer Science University of Arkansas, Fayetteville, Arkansas, USA · Department of Biomedical Engineering University of Arkansas, Fayetteville, Arkansas, USA
Interpretable Multi-Instance Learning Enables Early Prediction of Key Molecular Alterations from Routine Flow Cytometry in Acute Myeloid Leukemia
Sep 16, 2026Jonathan Legrand, Aguirre Mimoun, Baudouin Denis de Senneville +3Clinical Decision Support Interpretable ML
IMB, MONC · CHU Bordeaux · CHU Bordeaux, Inserm U1312 - BRIC +1

Mar 17, 2026J/K move · Enter open · S save

Understanding Cell Fate Decisions with Temporal Attention

Florian Bürger, Martim Dias Gomes, Adrián E. Granada, Noémie Moreau+1

Institute for Biomedical Informatics, Faculty of Medicine and University Hospital Cologne, University of Cologne, Cologne, Germany · Faculty of Mathematics and Natural Sciences, University of Cologne, Cologne, Germany · Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD), University of Cologne, Cologne, Germany+2

Microscopy Image Analysis Interpretable ML

Open paper AskPDF arXiv

Understanding non-genetic determinants of cell fate is critical for developing and improving cancer therapies, as genetically identical cells can exhibit divergent outcomes under the same treatment conditions. In this work, we present a deep learning approach for cell fate prediction from raw long-term live-cell recordings of cancer cell populations under chemotherapeutic treatment. Our Transformer model is trained to predict cell fate directly from raw image sequences, without relying on predefined morphological or molecular features. Beyond classification, we introduce a comprehensive explainability framework for interpreting the temporal and morphological cues guiding the model's predictions. We demonstrate that prediction of cell outcomes is possible based on the video only, our model achieves balanced accuracy of 0.94 and an F1-score of 0.93. Attention and masking experiments further indicate that the signal predictive of the cell fate is not uniquely located in the final frames of a cell trajectory, as reliable predictions are possible up to 10 h before the event. Our analysis reveals distinct temporal distribution of predictive information in the mitotic and apoptotic sequences, as well as the role of cell morphology and p53 signaling in determining cell outcomes. Together, these findings demonstrate that attention-based temporal models enable accurate cell fate prediction while providing biologically interpretable insights into non-genetic determinants of cellular decision-making. The code is available at https://github.com/bozeklab/Cell-Fate-Prediction.

Abstract

Toxicity Detection

Explore similar work

Understanding Cell Fate Decisions with Temporal Attention

Detection of extremely rare dead cells in fluorescence microscopy images

Interpretable Multi-Instance Learning Enables Early Prediction of Key Molecular Alterations from Routine Flow Cytometry in Acute Myeloid Leukemia