Microsoft Word Kurzweil, Ray The Singularity Is Near doc

Medicine.
If you obtain an electrocardiogram (ECG) your doctor is likely to receive an automated diagnosis using 
pattern recognition applied to ECG recordings. My own company (Kurzweil Technologies) is working with United 
Therapeutics to develop a new generation of automated ECG analysis for long-term unobtrusive monitoring (via 
sensors embedded in clothing and wireless communication using a cell phone) of the early warning signs of heart 
disease.
189
Other pattern-recognition systems are used to diagnose a variety of imaging data. 
Every major drug developer is using AI programs to do pattern recognition and intelligent data mining in the 
development of new drug therapies. For example SRI International is building flexible knowledge bases that encode 
everything we know about a dozen disease agents, including tuberculosis and H. pylori (the bacteria that cause 
ulcers).
190
The goal is to apply intelligent datamining tools (software that can search for new relationships in data) to 
find new ways to kill or disrupt the metabolisms of these pathogens. 
Similar systems are being applied to performing the automatic discovery of new therapies for other diseases, as 
well as understanding the function of genes and their roles in disease.
191
For example Abbott Laboratories claims that 
six human researchers in one of its new labs equipped with AI-based robotic and data-analysis systems are able to 
match the results of two hundred scientists in its older drug-development labs.
192
Men with elevated prostate-specific antigen (PSA) levels typically undergo surgical biopsy, but about 75 percent 
of these men do not have prostate cancer. A new test, based on pattern recognition of proteins in the blood, would 
reduce this false positive rate to about 29 percent.
193
The test is based on an AI program designed by Correlogic 
Systems in Bethesda, Maryland, and the accuracy is expected to improve further with continued development. 
Pattern recognition applied to protein patterns has also been used in the detection of ovarian cancer. The best 
contemporary test for ovarian cancer, called CA-125, employed in combination with ultrasound, misses almost all 
early-stage tumors. "By the time it is now diagnosed, ovarian cancer is too often deadly," says Emanuel Petricoin III, 
codirector of the Clinical Proteomics Program run by the FDA and the National Cancer Institute. Petricoin is the lead 
developer of a new AI-based test looking for unique patterns of proteins found only in the presence of cancer. In an 
evaluation involving hundreds of blood samples, the test was, according to Petricoin, "an astonishing 100% accurate in 
detecting cancer, even at the earliest stages."
194
About 10 percent of all Pap-smear slides in the United States are analyzed by a self-learning AI program called 
FocalPoint, developed by TriPath Imaging. The developers started out by interviewing pathologists on the criteria they 
use. The AI system then continued to learn by watching expert pathologists. Only the best human diagnosticians were 
allowed to be observed by the program. "That's the advantage of an expert system," explains Bob Schmidt, Tri-Path's 
technical product manager. "It allows you to replicate your very best people." 
Ohio State University Health System has developed a computerized physician order-entry (CPOE) system based 
on an expert system with extensive knowledge across multiple specialties.
195
The system automatically checks every 
order for possible allergies in the patient, drug interactions, duplications, drug restrictions, dosing guidelines, and 
appropriateness given information about the patient from the hospital's laboratory and radiology departments.