STATISTICAL DESIGN OF EXPERIMENTS AIDED BY ARTIFICIAL INTELLIGENCE

QUANTECH PROPOSES, DURING THE PHASE I SBIR, TO DETERMINE IF THEY CAN ASSIST IN THE STATISTICAL DESIGN OF ENVIRONMENTAL, INDUSTRIAL, AND BIOMEDICAL EXPERIMENTS BY UNITING ARTIFICIALINTELLIGENCE AND NUMBERICAL COMPUTATION ON A MICRO-COMPUTER. THEY WILL ATTEMPT TO CAPTURE THE INTERACTION BETWEEN STATISTICIAN AND CLIENT CONCERNING THE INTENT AND STRUCTURE OF AN EXPERIMENTAL STUDY, TO REPRESENT THIS MATERIAL AS AN AI KNOWLEDGE BASE, AND TO LINK AI ADVISORY FUNCTIONS WITH THE NUMERICAL COMPUTATION NEEDED TO MAKE SPECIFIC DESIGN CHOICES. SUCH A HYBRID SYSTEM WOULD PROVIDE GUIDANCE AND COMPUTATION CONCERNING SAMPLE SIZE, THEPOWER TO DETECT TRUE EXPERIMENTAL EFFECTS, AND THE MINIMAL DETECTABLE EFFECT. PHASE I OBJECTIVES INCLUDE SELECTING THE MAJOR STATISTICAL DESIGNS AND THE COMPUTATIONS WHICH LINK THE DESIGN FEATURES MENTIONED ABOVE, CREATING A SMALL KNOWLEDGE BASE AND COMPUTATIONAL MODULE RELEVANT TO STATISTICAL DESIGN, AND LINKING THEM BY A CONVENIENT AND INFORMATIVE USER INTERFACE.UNLIKE THE WIDELY ACKNOWLEDGED NEED FOR EXPERT STATISTICAL ANALYSIS OF EXPERIMENTAL DATA, LITTLE STRESS IS PLACED UPON APPROPRIATE STATISTICAL DESIGN OF EXPERIMENTS. THE CONSEQUENCE MAY BE A WASTE OF EXPERIMENTAL RESOURCES OR UNINTERPRETABLE EXPERIMENTAL DATA. THIS RESEARCH ADDRESSES A NEED OF ENVIRONMENTAL SCIENTISTS, ENGINEERS, AND THE PHARMACEUTICAL INDUSTRY FOR AN EVER-AVAILABLE, EXPERT STATISTICIAN TO ENSURE THE CREDIBILITY AND RELIABILITY OF PROJECTS THAT ARE STRUCTURED AS A SCIENTIFIC EXPERIMENT.