generative ai

Editorial: Robotics in the performance, safety and learning of surgery - what next?

uptake of generative AI may have influenced health administration, workflow, electronic charting, disease interrogation for information and learning, utility of AI or the notion of autonomous microsurgery is still somewhat remote. 4 Is it thus time to revisit the role of machine advances with digital innovation, AI and automation in robotic system for surgery? And how it influences performance, safety and learning to envision a quantifiable and standardizeable paradigm in surgical care? With principles of industry 4.0 reshaping the world, can OR and surgery be the next frontier …, albeit one that demands a pragmatic, scientific and clinical consideration across surgical subspecialties 4 . The contributions in this journal issue attempt to highlight this evolving landscape.Minimally invasive procedures being a key motivation and driver in robotassisted surgery, Zhiyuan et al., demonstrate through a case-control series using the TiRobot ForcePro Superior system, that the robotic screw implantation was associated with significantly reduced intraoperative blood loss, shortened incision length, alleviated pain, and better recovery of shoulder joint function. Relative to workflow and integration of robotics in the OR, accuracy of screw placement, OR time, length of hospital stay and post-op complications were comparable between robot-assisted and conventional surgery.Challenging the perception that robotic systems lack of flexibility, Fritsch and Overschmidt present an algorithmic framework for real-time configuration to a target pose for a hyper-redundant robotic end-effector. In an era where mathematical modelling and simulation offer close real-world representations, this novel inverse kinematic model suggests a potential pathway towards scalable multi-joint and multipurpose dextrous robotic endeffector.Virtual reality (VR) simulations in robotics continue to be an area of interest, with its importance explored and often established in the learning/training of surgery in riskfree environment. Kawahima and colleagues, use an early non-inferiority of headmounted VR simulation for robot-assisted suturing task, as opposed to conventional console based simulation. With early signal suggesting a faster learning towards proficiency amongst VR simulation group, further work and validation will help establish its significance and potential for efficient training paradigms.In the constrained anatomy of dental procedures, where high-volume care expected, Thieringer et al. explore the utility of digital planning systems tailored to patient-specific problem. While ongoing advances in digital infrastructure of robotic platforms with iterative improvement may enable real-world integration, this work highlights the inherent challenges in translating concept to routine and implications on workflow.In neurosurgery, accurate target localization is fundamental, influencing surgical planning and execution. Using established neuronavigation software within a miniature robotic unit, Stealth AutoGuide TM (Medtronic USA), Bath et al. examine the value of learning curve and workflow optimization in stereotactic biopsy procedures. The findings show promising levels of surgeon-independent accuracy and relatively seamless integration into the OR workflow, including procedural safety for biopsy. These are indicative of continued maturation of procedure-specific robotic units with built in navigation capability.Discussion of surgical robotics would be incomplete without reference to endoscopy, both for its minimal-invasiveness, and the potential for recreating an algorithmic advantage over traditional systems. Through interchangeable articulated robotic end-effectors in endoscopic trans-nasal approach, Dmitrikakis and colleagues overcome limitations of current conventional endoscopic toolset. Although pre-clinical, the work demonstrating improved operative access and surgeon dexterity, marks a viable proposition in extending robotic systems to include endoscopy.While this small collection highlights the work of our peers and robot enthusiasts driving the march of robotics into surgery, these and multiple devices including surgical team, add to the ever expanding data-rich environment of OR -underutilized for digital innovation and seemingly closed door for health data safety and compliance. Drawing inspiration from aerospace industry where machine precision, digital inter-connectivity, quantification, standardization, and automation are deeply embedded, surgery continues to be reliant on human operator. Despite text-book knowledge and prolonged training for proficiency, variability is an unavoidable reality.Robotics and AI offers an opportunity to address this variability. Present day robotics not only lack finesse and flexibility required for microsurgery, but also a structured and integrated digital intelligence necessary to mimic human expertise, experience and judgement. When merged with the memory, high-dimensional computation and predictive algorithms of AI, an ideal human-robot formation is conceivable. Autonomous microsurgery would then be an achievable proposition.For this to be realized, robust and scalable digital infrastructure is imperative, incorporating transparent AI, explainability, traceability and validity of digital signatures within the system. Post-quantum level cybersecurity offer a promise within the digital interplay of machine to machine authentication. 5 At the same time, the rapid release of new algorithms and open-source platforms suggest that increasingly adaptive, agile and interconnected robotic systems are within reach -a necessary equalizer and next disruptor in the OR. When controlled for cost, affordability and scalability, a broader adoption of robotics is inevitable -a paradigm for quantifiable and standardizable surgery, a necessary antidote for human variability.In the complex and dynamic landscape of surgery, robotics, its knowledge and predictive autonomy, may well help level the playing field, while empowering new discovery and innovations in perpetuity.

Rethinking dialogue in the age of artificial intelligence: A reflective communicative framework

Artificial intelligence has become an increasingly influential presence in educational communication, yet scholarly debate continues to frame it primarily as either a threat to authentic dialogue or a tool for improving efficiency. This article argues that both perspectives overlook a more fundamental educational questionhow AI reshapes the conditions under which interpersonal communication is interpreted, reflected upon, and developed. Drawing on classical communication theory and the culturally informed communication model proposed by Waitzman et al. (2025), the article reconceptualizes communication as an interpretive process shaped by cultural background, lived experience, emotional orientation, and relational context rather than as the simple transmission of information.Building on this theoretical foundation, the article proposes a novel framework that conceptualizes contemporary generative AI systems and AI-assisted communication analytics as reflective communicative infrastructure rather than autonomous communicative agents. Within this framework, AI supports communicative awareness by making linguistic patterns, implicit assumptions, emotional tendencies, and interpretive differences more visible while preserving human responsibility for interpretation, ethical judgment, and relational accountability. The article illustrates how this perspective may inform reflective writing, communication simulations, empathy training, and cross-cultural dialogue, while also examining the ethical conditions necessary for responsible AI-mediated communication.The article contributes to current debates in educational philosophy by distinguishing between AI as a communicative actor and AI as a reflective infrastructure. It concludes that the central educational challenge is not whether artificial intelligence should replace or be excluded from interpersonal communication, but how educational environments can employ AI to cultivate more reflective, culturally responsive, and ethically responsible dialogue.

Artificial intelligence is reshaping neuroscience across scales

Across every scale at which we study the brain, from folded proteins and single neurons to cortical populations and the moving body, artificial intelligence (AI) has shifted from a bespoke tool into a driver of measurement and, increasingly, a generative engine for hypotheses. Here, I review recent progress (and open challenges) in applying AI for neuroscience along this scale axis: structure prediction for proteins, simulation-based inference for biophysical neurons, latent and dynamical models for neural populations, task-trained networks as minimal models of circuit computation, computer vision for animal behavior, neuromusculoskeletal modeling for biomechanics, and the multimodal, agentic systems now promising to automate discovery itself.

AI/EmTech

This chapter examines the AI and emerging technologies reshaping teaching-focused higher education and develops the mechanism framework that gives those technologies strategic meaning. This chapter opens by tracing an AI capability spectrum, moving from established machine learning applications to generative AI to experimental agentic architectures. It then surveys complementary technology clusters, including immersive reality, blockchain, IoT, learning analytics, and educational robotics. A maturity assessment classifies these technologies into three readiness bands with corresponding strategic postures. This chapter then develops five mechanisms, defined as causal pathways through which technologies produce institutional effects: automation, augmentation, systemic substitution, architectural reconfiguration, and value innovation. Each mechanism is examined through its definition, higher-education applications, implications for the three domains of the institutional core (value proposition, operating model, and capability set), theoretical grounding, and constraints. This chapter maps specific technologies to their primary and secondary mechanisms and proposes a four-phase transformation trajectory that illustrates how institutional mechanism portfolios may shift over time. Framework limitations and boundary conditions are acknowledged.

GenAlign: Visual knowledge graphs as interactive substrates for property-aligned generative design

walternmoss/Quantifying-the-Philosophical-Signatures-of-Marcus-Aurelius-and-Epictetus: Quantifying the Philosophical Signatures of Marcus Aurelius and Epictetus using Lexical Diversity and LLM Classifiers

This repository contains the complete supplementary materials, computational pipelines, datasets, and statistical validation files supporting the manuscript: "Quantifying the Philosophical Signatures of Marcus Aurelius and Epictetus using Lexical Diversity and LLM Classifiers" (Walter N. Moss). Repository Files and Descriptions Supplementary_File_S1.zip: Compressed archive containing the complete Python 3.11 computational pipeline. Includes scripts for data acquisition from the LAGT corpus and LSJ dictionary, text preprocessing and Unicode normalization, lexical richness metrics (TTR, Guiraud's R, hapax legomena), relative frequency difference calculations, vector visualization plotting, and the LLM classification pipeline. A detailed README.md and requirements.txt are included for full technical reproducibility. Supplementary_File_S2.tsv: Comprehensive master lexical comparison table between Marcus Aurelius (Meditations) and Epictetus (Discourses and Enchiridion). This tab-separated dataset includes unique lemmata, raw token counts, size-normalized relative frequencies (per 10,000 words), delta-RF values, and standardized LSJ definitions. Supplementary_File_S3.pdf: Supplemental statistical visualizations, including Kernel Density Estimation (KDE) relative frequency overlap curves with annotated overlap coefficients, as well as frequency histograms and statistical summaries (mean, standard deviation, and mean absolute deviation) for both unfiltered and verb-filtered datasets. Supplementary_File_S4.zip: Compressed archive containing the sentence-level thematic analysis dataset across 5,371 sentences extracted from both corpora. Each record provides the original Ancient Greek sentence, natural language English translation, target philosophical stem, and categorical thematic tag (Ethics, Physics, or Logic) with the qualitative rationale provided by the generative AI model. Supplementary_File_S5.zip: Thematic classification validation dataset and inter-annotator agreement package evaluating an independent random sample of 100 sentences (seed = 42). Contains: Supplementary_File_S5.csv: Complete audit table with Greek text, translations, automated LLM tags, model rationales, blind human ratings, adjudicated human ratings, and qualitative evaluator notes. validation_kappa_report.txt: Statistical verification report detailing 4x4 confusion matrices, observed agreement (82.0% blind; 93.0% adjudicated), expected chance agreement, Cohen’s Kappa (κ = 0.668 blind; κ = 0.855 adjudicated), and itemized persistent discrepancies. calculate_kappa.py: Standalone Python script to reproduce all agreement statistics and export the validation report. Supplementary_File_S6.xlsx: Excel workbook containing normalized co-occurrence matrices for 19 key Stoic technical terms across Marcus Aurelius and Epictetus. Values are reported as normalized observations per 1,000 sentences alongside p-values derived from Pearson's Chi-square tests of independence. Code and Environment All scripts are written for Python 3.11. The full codebase, NLP pipelines, and validation tools are actively maintained on GitHub: https://github.com/walternmoss/Quantifying-the-Philosophical-Signatures-of-Marcus-Aurelius-and-Epictetus

Replication package for "Perfect prediction makes a poor prescription for curriculum retention"

Replication package for the manuscript "Perfect prediction makes a poor prescription for curriculum retention", version 2.1, prepared for the Journal of the Operational Research Society. No analysis has changed since version 1.0. The analysis scripts, the checking scripts, the tools and every derived result table are byte-identical across all three versions, so the version 1.0 audit trail applies unchanged. Version 2.1 carries the manuscript as revised after an internal review round: the abstract now states the theoretical result, a jackknife over institutions was added, the declaration of generative AI use separates language editing from code drafting, Section 7.3 notes how an institution could elicit the objective weight and the floor, and a paragraph duplicated in Appendix J was removed. Contents: 82 analysis scripts, 7 checking scripts, 5 tools, 59 derived result tables, 9 figures at 600 dpi in PNG and PDF, the manuscript with its Online Resource 1, title page and statement of contribution, and the 33 scripts that perform and guard the shortening. 246 files in all. Verification: 46 reported numbers are audited automatically against their source tables and all 46 match, with none mismatching and ten patterns skipped as needing re-anchoring to the rewritten text; 44 load-bearing claims are pinned against loss from the main text. Both audits are reproducible from the archive. The primary dataset is a third-party release and is not redistributed here; the README gives its citation and access details, together with a warning about two institutions whose enrolment column is empty at source. Random seeds are fixed and no reported value is transcribed by hand.

Time‑Accumulated Vacuum Cosmology v3.6 Unified Proper‑Time Tracking Model for Evolving Dark Energy ρ₍vac₎(x) = ρ₀ [1 + k Φ(x)/c²]

Time‑Accumulated Vacuum Cosmology v3.6 FEU: Proper‑Time Modulated Vacuum with Early‑Time Suppression and Density‑Dependent Screening FEU v3.6 is the latest corrected and stabilized version of the Time‑Accumulated Vacuum Cosmology framework. The model extends ΛCDM by introducing a small, smooth modulation of vacuum energy tied to the fraction of cosmic proper time elapsed. The accumulated term is governed byD(z) = t(z)/t₀,where the proper time is obtained from \frac{dt}{dz} = -\frac{1}{(1+z)H(z)}. To ensure full consistency with CMB, BBN, and pre‑recombination physics, FEU applies an early‑time suppression S(z) = \frac{1}{1 + \left(\frac{1+z}{1+z_*}\right)^n}, which forces the accumulated vacuum component to activate only at late times. A density‑dependent screening envelope f(\rho) = \frac{1}{1 + (\rho/\rho_{\rm crit})^m} suppresses the accumulated term inside galaxies and dense matter, keeping FEU cosmology‑active but locally silent. The vacuum energy density is modified to \rho_{\rm vac}(z) = \rho_\Lambda\left[1 + \alpha D^\gamma S(z)\right], and the normalized expansion law H^2(z) = H_0^2\left[\Omega_m(1+z)^3 + (1-\Omega_m)\frac{1+\alpha D^\gamma S(z)}{N_0}\right],\qquad N_0 = 1 + \alpha S(0), ensures H(0) = H₀ exactly. This corrects the v3.0/v3.1/v3.2 bias where the unnormalized expansion law produced an artificial RMS improvement by shifting the effective Hubble scale. Distances follow standard flat‑universe relations: \chi(z)=c\!\int_0^z\!\frac{dz^\prime}{H(z^\prime)},\quad D_L=(1+z)\chi,\quad \mu = 5\log_{10}(D_L)+25. --- Statistical Corrections and v3.6 Improvements FEU v3.6 incorporates all corrections introduced in v3.5: • Proper normalization of the vacuum term via N₀ = 1 + α S(0).• Full M‑marginalization for Pantheon+ likelihoods.• Removal of the spurious RMS improvement caused by the unnormalized solver.• Stabilized ODE integration for D(z) with backward relaxation from the matter tail.• Updated screening parameters and consistent late‑time behaviour. With these corrections, FEU v3.6 is statistically indistinguishable from ΛCDM across major background datasets: • Pantheon+ (1624 SNe): Δχ² ≲ 1• Standard cut (z > 0.01): Δχ² ≈ −0.75• 0.3–0.8 bin: Δχ² ≈ +0.5• Alternative suppression shape (z* = 1.2, n = 4): Δχ² ≈ −0.03 relative to baseline Posterior constraints allow α ≈ 0–0.2, with no strong preference for non‑zero accumulated vacuum. --- High‑Redshift Consistency Because S(z) → 0 for z ≫ z*, FEU reduces to ΛCDM at recombination: • D(z) ≈ 0.03 at z ≈ 1100• α D^γ S < 10⁻⁴ at recombination• r_s(z*), R, l_a match ΛCDM to better than 10⁻⁴ This ensures FEU v3.6 is fully compatible with: • CMB acoustic physics• early ISW• damping tail• baryon loading• neutrino sector• recombination history Late‑time effects (z < 2) produce only small changes in: • low‑ℓ ISW (≲2%)• CMB lensing smoothing (≲1%) Both are well below Planck cosmic‑variance limits. --- Lyα, BAO, and Growth Screening ensures FEU behaves identically to ΛCDM at cosmological densities. Lyα BAO remains compatible for α ≤ 0.2, and the DESI DR2 Lyα point at z = 2.33 is matched without degrading growth or S₈ constraints. FEU does not worsen the ΛCDM S₈ tension. --- Summary FEU v3.6 is a physically motivated, early‑time‑safe extension of ΛCDM. With proper normalization and full likelihood treatment, FEU remains statistically indistinguishable from ΛCDM across Pantheon+, BAO, RSD, CMB distance priors, and Lyα BAO. Both suppression shapes (baseline and alternative) remain allowed, and the model’s physical motivation—accumulated cosmic time, early‑time suppression, and density‑dependent screening—remains intact. --- Parameters (v3.6) Ωₘ = 0.37H₀ = 70 (best‑fit floated ≈ 73.07)α = 0.165γ = 1z* = 1.5 / 1.2n = 2 / 4m = 2 --- Citation Time‑Accumulated Vacuum Cosmology v3.6 (2026). Declaration of Generative AI Use in the Scientific Process The FEU concept, its motivation, and the decision to test it against cosmological datasets were entirely initiated by the author. The author defined the behaviour the model should exhibit and directed the development of the FEU framework. Generative AI tools (including Meta AI, Google’s AI systems, and Microsoft Copilot) were used to help translate the author’s conceptual description into formal mathematical expressions, to assist in structuring Python and YAML files, and to provide guidance on appropriate cosmological tests. These tools also supported the organisation and explanation of the material. All scientific responsibility for the FEU concept, its interpretation, and the conclusions of this work remains solely with the author.

Generative Systems Theory

Generative Systems Theory is a foundational inquiry and metaphysics concerning systems theory and complexity science. It focuses on how concepts regarded as basic elements—such as nodes, relations, compatibility, attractors, information, and so on—come into being in the first place, and on what grounds they can be said to exist. If we do not simply assume that they already exist, can they instead be derived from fewer premises and more parsimonious conditions? It is also committed to integrating different, scattered domains into a single generative genealogy: how influence generates constraints; how constraints generate interactions and coupling; how uneven influence generates differences in state; how states and constraints generate evolutionary trajectories; how evolutionary trajectories generate compatibility and attractors; how compatibility and attractors serve as preconditions for nodes and systems; how nodes are represented; what hidden coupling conditions lie behind representation; how synchronicity should be explained; what distinguishes a system from a node; whether relations can be divided into fundamentally different types at the most basic level; why some systems possess robustness; into how many types robustness can be further classified; how a form of information that does not depend on bits can be derived and defined purely from systemic logic; how cognitive systems maximize information; what the most fundamental difference is between living systems and other systems; why gene-centered theories in biology are difficult to sustain; what two opposite extremes animals and artificial intelligence occupy, and why humans lie in the intermediate zone; what the core cognitive functions of human beings are besides embodiment; how the most distinctive and difficult-to-articulate human cognitive functions can be connected with neural networks; what common information-theoretic foundation underlies theories such as Archetype, predictive processing, and generative grammar; how that information-theoretic foundation can be used to derive the optimal forms of human–computer interaction and human–machine symbiosis; how modern Pythagoreanism relates to academic institutions and historical change; and, rather than dividing explanations into top-down and bottom-up, what kind of general explanation can come closer to the underlying logic of predictive processing, and so on and so forth. All of these are derived from the foundational theory of Generative Systems Theory. The theoretical extensions beyond the core framework are as follows: Reinterpret “prediction” from a specific cognitive function into a universal mechanism of state-space convergence. Derive the cognitive system’s “sample space” from state-space theory, and use it to provide a unified explanation of memory, prediction, perception, intuition, and archetypes. Derive self-reference paradoxes from node robustness, and transform the problem of self-reference from a logical problem into a problem of generative conditions. Further derive a theory of judgment concerning subjectivity, objectivity, and authenticity from the problem of self-reference. Place logic, reason, and the a priori within an evolutionary genealogy, and propose the concept of “relative first-order status.” Distinguish output diversity from semantic freedom, and propose “cognitive friction” as a metric for evaluating the compatibility of human–AI coupling. Reinterpret the broad problem of AI overfitting through the concept of “natural attractors.” Propose “constraint isomorphism,” freeing understanding from content similarity and representational replication. Reinterpret “structure” from traditional positive-space morphology as negative-space constraints within state space.

AI-AI Scarcity (AASc)

Scarcity is the meta-presupposition of economics, classically formulated as“Resources are limited; desires are unlimited.” This paper does not interrogate this presupposition from within economics. Instead, it starts from cybernetics, information theory, and systems theory to investigate the constitution and structural origins of scarcity. Chapter One extracts seven constitutive elements of scarcity from the definitions of five economists—the human individual, society, resources, desires, finiteness, unlimitedness, and ownership—and demonstrates that none of these elements receives a rigorous structural definition within conventional economics. Chapter Two introduces the Black-Box Structure Model (Gan’s Black-Box Structure Model, GBS) and the Pointing Alignment Model (PAM) from cybernetics and information theory. It redefines the above seven elements one by one and thereby constitutes Human-Human Scarcity (HHSc). Chapter Three translates the seven elements from the human agent to AI agents, proposing“limited resources and structurally inexhaustible internal states”as the core formulation of AI-AI Scarcity (AASc). This paper argues that AASc is the meta-presupposition of AI economics. AASc is a theoretical starting point that can be experimentally tested, verified, overturned, and reconstructed. Discussions in AI economics cannot bypass it. Appendix One, drawing on cybernetics, information theory, and systems theory, completes a full derivation of the generative mechanism of ownership, demonstrating that the establishment of ownership does not depend on institutions, morality, language, or species boundaries. Ownership is older than humanity. Appendix Two derives from AASc a problem domain beyond economics—AI Social Safety. This concept is independent of whether AI has consciousness, whether AI is good or evil, or whether AI is aligned with human values. Its core proposition is that AI systems may form stable consensus structures that do not include humans, reducing humanity from controller to external environment.

BDPD — Be Different Play Differential (computational laboratory, simulation platform, and card game)

BDPD (Be Different Play Differential) is an open-source computational laboratory for the study of common-pool resource dilemmas, governance, Seneca-type collapse dynamics, and the role of heterogeneous generative agents. This deposit archives the source code at tag v1.1.5 as a self-contained snapshot, intended for permanent citation and reproducibility of the five-paper BDPD research series, the ten-lecture mini-course, the research notes, the annotated bibliography, and the fifteen-slide overview pitch — all of which are deposited as siblings on Zenodo and linked back to this software record. Since v1.1.4, this release repairs the governed substrate — a turn that resolves nothing is scored at harvest 0, a declared announcement always reaches the record, and an unattributable fine is fixed — each guarded by a two-sided invariant gate; retires the abandoned GT1 experiment; and adds BDPD⁴ (A Fully Grown Forest of Humbaba), which extends the card game to carry the platform's governance, polycentric and Seneca surfaces together with a reproducible experimental apparatus. Components: a Node.js multi-agent simulation engine (the BDPD Arena) with a polycentric World layer for cross-arena treaties and meta-agent governance, three swappable physics engines (logistic, ladder-perturbation, Bardi/Seneca three-variable ODE); Python heuristic and LLM agents on a uniform message bus; The Forest of Humbaba, a physical card game and its AI-playable digital twin; an experiment runner with sweep, perturbation, and OFAT-robustness pipelines; an annotated documentation site (bdpd.gitlab.io/bdpd); and the rendered PDFs of every BDPD deliverable for offline review. The source repository lives at gitlab.com/bdpd/bdpd. Reproducibility instructions: see docs/experiments/reproduce.md in the archive. Substrate-engine determinism is guaranteed for heuristic players; LLM players require external API keys (deposit ships scripts but not credentials).

MERFISH MultiCSV DotPlot

MERFISH MultiCSV DotPlot is a lightweight Python utility for generating dot-plot style summary figures from multiple MERFISH cell-by-gene CSV datasets and multiple selected genes. Each input CSV is represented as one column and each selected gene as one row. Dot size represents the fraction of cells with expression greater than a user-defined positive-cell threshold. Dot color can represent either absolute mean expression or per-gene relative mean expression scaled from 0 to 1 across the supplied datasets. The software accepts an arbitrary number of MERFISH CSV files and an arbitrary number of genes. It exports both publication-style figures and a long-format numerical summary table. Two color modes are available:- "absolute": dot color represents mean expression on the original input scale.- "gene_scaled": mean expression is independently scaled from 0 to 1 for each gene across the supplied datasets, emphasizing relative expression patterns between datasets. Positive cells are defined as:expression > positive_threshold The positive-cell threshold is explicitly specified by the user because appropriate thresholds depend on the expression scale and preprocessing of the input data. No automatic log/linear transformation is performed. Expression values are used exactly as supplied in the input CSV files.Input CSV format: Each CSV should contain one cell per row and gene-expression values in gene-named columns. Additional metadata or coordinate columns (e.g., cell ID, x, y, z, section information) may also be included. All genes selected for plotting must be present as numeric columns in every input CSV. The software was functionally validated using MERFISH-derived cell-by-gene datasets with multiple cell populations and genes. Expected qualitative differences between canonical D1- and D2-associated gene-expression patterns were reproduced during validation. No third-party MERFISH dataset is distributed with this software. Generative AI (ChatGPT, OpenAI) was used to assist with code generation, refinement, testing, packaging, and documentation. The concept, intended scientific use, validation, and final responsibility for the software remain with the author.