Laboratory of High-Dimensional Causal Inference and Big Data Structural Analysis

Overview

Our laboratory focuses on developing and applying data-analysis methods based on causal inference for large-scale, high-dimensional data obtained from medicine, environment, economics, and related fields.

While big data analytics has traditionally emphasized improving predictive accuracy, real-world implementation and policy decision-making also require appropriate evaluation of cause–effect relationships and the impact of interventions. Building on statistics, machine learning, and mathematical modeling, our research emphasizes analyses that go beyond simple correlation and address data as complex systems from a causal perspective.

We position concepts from quantum information theory as an educational framework for understanding notions such as state spaces and observational constraints, helping students develop essential mathematical literacy for complex data structures. At the same time, the core of our research lies in developing and validating causal inference methods that are feasible on current classical computing environments.

Research characteristics

• Causal inference tailored to high-dimensional, nonlinear data with missing values

• Statistical methods for estimating intervention effects from observational data

• Methodological validation through applications in medicine, environment, and economics

• Integrated use of computational statistics, machine learning, and causal inference

• Large-scale analysis using high-performance computing with a focus on reproducibility

Research objectives

• Development and advancement of causal inference methods applicable to real-world data

• Supporting data-driven decision making in medicine, environment, and economics

• Fostering human resources capable of understanding and implementing high-dimensional data analysis and causal inference

We organize the well-known “5 V’s” of big data—Volume, Variety, Velocity, Veracity, and Value—from the viewpoint of causal inference and explore appropriate analytical methodologies.

Main research topics

Medical big data and causal inference
We analyze causal relationships among psychiatric disorders, dental diseases, and lifestyle-related diseases using large-scale datasets such as insurance claims and health checkup records. Disease progression and treatment effects are evaluated not merely by prediction but through intervention effect estimation and counterfactual analysis, aiming to contribute to preventive medicine and health policy.

Causal structure analysis of environmental, climate, and agricultural data
Using sensor networks, drones, and meteorological data, we estimate the causal impact of environmental factors on ecosystems and agricultural production. We address nonlinear and nonequilibrium dynamics to contribute to the design of sustainable environmental and agricultural policies.

Economics, finance, and causal risk evaluation
By integrating medical, environmental, and economic data, we analyze how policy changes and market interventions affect risk structures. In financial data analysis, we emphasize causal evaluation of institutional and policy interventions rather than mere optimization or prediction.

Research environment and roadmap

Research is conducted using high-performance computing clusters in combination with cloud computing environments.
Quantum simulators and quantum computing platforms are positioned as auxiliary educational and conceptual tools, used for conceptual exploration and mathematical understanding.

• Phase 1 (current): Application of causal inference methods to real data and creation of social impact

• Phase 2 (3–5 years): Advancement of high-dimensional and complex-system causal models

• Phase 3 (5–10 years): Educational and theoretical exploration of new computational paradigms

Interdisciplinarity and international collaboration

We promote interdisciplinary research across medicine, environment, and finance, and we participate in collaborative projects with universities, research institutes, and industry partners, including international initiatives, to address societal challenges from multiple perspectives.

Prospective students

• Those who wish to solve social issues using big data

• Those interested in medicine, environment, or finance

• Those who want to learn statistics, machine learning, and programming

• Those eager to take on new technologies

• Those who wish to work in an international research environment

No specific prior knowledge is required. With motivation and curiosity, students will receive careful guidance starting from the fundamentals.

Laboratory policy

We aim to establish methodologies that contribute to the analysis of real-world data while clearly distinguishing prediction vs. explanation and correlation vs. causation. Through research grounded in realistic computing environments and real data, we pursue both academic insight and societal relevance.

Profile: Professor of Suwa University of Science, visiting associate professor of Division of Cardiovascular Medicine, Department of Internal Medicine, Kurume University School of Medicine.

Detailed achievements: Google Schalor, ORCID, Researchmap

Reference IDs: DA18461433, NDL: 00980363, VIAF ID: 256308865, WorldCat Identities: lccn-n2018006921, KAKEN: 60449238, Webcat Plus, Amazon

Contact: Kazuo Ishii, Ph.D.

Department of Applied Information Engineering, 

Faculty of Engineering, Suwa University of Science, 

5000-1 Toyohira, Chino-shi, Nagano 391-0292, JAPAN. 

Tel.(+81)-266-73-1201