Research Themes

Vision & Approach

Life is sustained by a delicate equilibrium between fluctuation and order.
Our laboratory uses the cardiac sarcomere—the smallest self‑oscillating unit of heart muscle—as a stage to pursue the essence of life dynamics across molecular, cellular, and tissue scales.
▶ Details

I. Chaordic Homeodynamics (Periodic Chaos Homeostasis)

Using our proprietary SL‑nanometry (3 nm spatial, 2 ms temporal resolution), we were the first to demonstrate heat‑induced self‑oscillations (HSOs) whose period remains constant while the amplitude varies chaotically. By recurrence plots, Lyapunov exponents, and mathematical modeling, we are quantifying how chaotic fluctuations in individual sarcomeres collectively generate a smooth, stable heartbeat—Chaordic Homeodynamics.

• Key papers
  Shintani SA. Biochem. Biophys. Res. Commun. 760, 151712 (2025) ▶ Details
  Shintani SA. Biophys. Physicobiol. 21, e210006 (2024) ▶ Details
  Shintani SA. Biochem. Biophys. Res. Commun. 611, 8‑13 (2022) ▶ Details
  Shintani SA et al. Sci. Rep. 10, 20468 (2020) ▶ Details
  Shintani SA et al. Biochem. Biophys. Res. Commun. 457(2), 165‑170 (2015) ▶ Details
  Shintani SA et al. J. Gen. Physiol. 143(4), 513‑524 (2014) ▶ Details

• Significance
  Addresses the fundamental question “Why can cardiac muscle beat without fatigue?”
  Explains how biological rhythms reconcile order and fluctuation (Chaordic Homeodynamics).
  Provides a new theoretical framework for arrhythmogenesis and dynamical diagnosis.

• Toward Societal Implementation
  We are exploring ultra‑early diagnostics for heart failure/arrhythmia based on rhythm “chaoticity,” and new design principles for next‑generation pacemakers.

II. Biomedical Engineering Applications & Social Implementation

Techniques born from sarcomere studies—temperature‑field control, nanometer imaging, AI analytics—are powering diverse collaborations:

• Titanium bone‑repair materials (antibacterial & bioactive coatings) ▶ Details

• Non‑invasive vital sensing (AI + optical methods) ▶ Details

• Cryo‑flash‑freezing & cryo‑EM for dynamic molecular imaging ▶ Details

We actively partner with universities, companies, and medical institutions to create new technologies in medicine, healthcare, and biomaterials.

III. Development of Measurement/Instrumentation & Analysis Methods, and the AI Utilization Guideline (AI2L)

To accelerate basic science, we develop custom microscopy systems, temperature‑control (heating) systems, and real‑time live‑imaging techniques for electron microscopy.
In parallel, we propose AI to Learn (AI2L)—an operational guideline that confines AI to a learning‑support role and removes any black‑box components from the final deliverables (preprint prior to peer review).

We are also studying arithmetic generalization using minimal GPTs. Even when exhaustive 2-digit addition already saturates the local digit-transition rules needed for later addition, 3-digit failure does not disappear; instead, it unfolds through staged computational barriers: a layout barrier, a carry-flag semantic error at the hundreds position, a conditional recomposition bottleneck, and a late tens residual. This line of work aims to replace a single held-out score with an auditable, experimentally separable account of what actually fails in arithmetic OOD generalization.

AI2L: Four Pillars

• Human sovereignty over final decisions

• Accountability ensured through human verification

• Privacy/Security with information protection placed first

• Green AI for energy efficiency and long‑term sustainability

Examples

• Use Grad‑CAM and related XAI tools to externalize tacit knowledge and distill it into human‑understandable rules

• Enumerate candidate laws via symbolic regression, followed by human theory‑building (contributing to the deepening of the Chaordic concept)

• Audit and slim AI‑generated code so it becomes a human‑owned, lightweight asset

• Reduce information‑leak risk during cloud use through reversible anonymization, etc.

Related material (preprint)

• Shintani SA. AI to Learn (AI2L): Guidelines and Practice for Human‑Centered AI Utilization as a Learning Support Tool — Four Pillars of Black‑Box Elimination, Accountability, Information Protection, and Energy Efficiency. Jxiv (preprint), DOI: 10.51094/jxiv.1435 ▶ AI2L Details

• Shintani SA. Arithmetic OOD Failure Unfolds in Stages in Minimal GPTs. arXiv (preprint), DOI: 10.48550/arXiv.2603.26828 ▶ Details

Note: These preprints have not undergone peer review; their contents may be updated or revised.
 

IV. Diverse Challenges & Research Styles

• Cryo‑EM exploration (flash‑freezing × dynamic molecular analysis) ▶ Details

• Fusion of AI, mathematics, and physiology (AI2L, Grad‑CAM, symbolic regression) ▶ Details

• Promotion of international collaborations, conference outreach, and academia–industry projects ▶ Details

• Fostering young talents and next‑generation researchers ▶ Details
  
  

We are a new interdisciplinary laboratory that unites biophysics, physiology, AI, and engineering to achieve both a fundamental understanding of life dynamics and tangible benefits for future medicine and society.