The microprocessor and medicine.

In November, 1971, Intel, a small start-up company, announced the first microprocessor--the Intel 4004. It used solid state integrated circuit technology that the firm had been using to make memory "chips" to develop the logic circuitry of a computer's central processing unit. A tiny company in the land of the gold rush, where most dreams of fortune never materialized, Intel was not likely to threaten the titans of computer manufacturing, such as IBM, which was growing at a torrid pace to meet the demand for its System 360 business computing systems.

Less than 25 years later, computing products made with the microprocessor progeny of the Intel 4004 would overwhelm IBM in the marketplace, and lead it to lose ignominiously more than $8 billion in a single year, and start a process to cut more than 100,000 employees from its work force. By then, most of IBM's product lines and computing platforms, from equipment for secretaries to its fastest supercomputers, had already embraced microprocessor technology, but the organization still thought of itself synonymously with big blue mainframes that the microprocessor had miniaturized.

Today, supercomputers are parallel processing machines in that they are made of parallel arrays of microprocessors. In 1994, an Intel Paragon supercomputer made of about 1,024 Pentium microprocessors became the world's fastest computer, achieving more than 280 billion floating point operations (gigaflops) per second. In September, 1995, Intel and Sandia National Laboratories announced a contract with the Department of Energy to create a Paragon supercomputer with more than 8,000 Pentium Pro microprocessors that will surpass 1.8 trillion floating point operations (1.8 teraflops) per second.

Recently, in the December, 1996, issue of Byte Magazine, the microprocessor was described as "the life-support system for the modern world." Not a bad performance for a device born only 25 years ago. Today, there is far more gold in Silicon Valley, south of San Francisco, than was ever found in the mountains east of Sacramento.

The pace of change

In the next five to 10 years, the pace of technological improvement in microprocessors will accelerate, and so will concomitant enhancements to the performance of computers that contain them. Intel and its competitors, including IBM, will continue to make more efficient microprocessors, and computer manufacturers will include parallel arrays of microprocessors in their equipment. Of the likelihood that technological improvements in microprocessors will decline, or stop, Mark Bohr, a research fellow at Intel, told Byte Magazine in April, 1996, "There's no sign of the technology slowing down. If we're going to run into a wall, it's more than 10 years out." The technological "wall" he refers to is a physical inability to condense more microscopic transistors onto microprocessor space.

In the same issue of Byte Magazine, John Kelly, a Vice President with IBM, said that "with CMOS (complementary metal-oxide semiconductor) technology and a lot of hard work, in a decade we'll be using X-ray lithography and other techniques to deliver a (micro)processor that has 50,000,000 to 100,000,000 transistors and operates at 1 gigahertz (1 billion cycles per second)." For comparison, the current Pentium Pro microprocessor includes 5.5 million transistors and operates at 200 megahertz (200 million cycles per second). That's a lot of room for improvement!

Not only are microprocessors improving in calculating speed, they are improving in multimedia capabilities. The MMX Pentium chip from Intel will be sold in early 1997 in versions with clock cycle speeds of 166 MHz and 200 MHz, optimized for multimedia computation, to handle high-resolution images, video, and high-fidelity sound more efficiently and effectively. The MMX improves video and audio performance two- to four-fold, and can render three-dimensional graphics that compare favorably with specialized graphics microprocessors. For routine processing of data files and numerical calculations, the MMX is about 10 percent faster than earlier Pentium chips with the same clock speeds. Improvements in processing video and audio signals will enhance the performance of computers with the MMX microprocessors in rendering multimedia content for education, entertainment, and video conferencing.

The Internet's World Wide Web is more demanding of multimedia processing than the text-based Internet and business networks used by most people today. But the World Wide Web is the future. The MMX microprocessor was designed for computer games, such as Quake, that involve high-resolution graphics and sound, and for multimedia education and conferencing. The personal computer is becoming the window to the multimedia digital world of cyberspace. Microprocessors will continue to improve from one generation to the next to make cyberspace as realistic as possible--and with their improving capabilities will come myriad applications for health care services, either not cost-effective, or not even contemplated, until now.

Microelectronics are improving faster than our expectations

We need to drop our preconceived notions about how much computing power and communication capability cost. Those expenses will continue to decline as new features and functions are added to personal computers, and they become lighter, more portable, more communicative, and more capacious for storing digital data. For instance, in 1974, the Intel 8080 microprocessor delivered one million instructions per second (1 MIPS) for $1,250. Today, Intel's Pentium Pro microprocessor delivers 1 MIPS for about $1. The cost of computing, estimated in the price of a million instructions per second, has fallen by more than three orders of magnitude in 22 years.

In terms of memory storage, the price of storing one thousand bytes of data in computer random access memory has fallen in the last 12 years from $15 per kilobyte to less than one cent per kilobyte. In terms of permanent magnetic data storage, the price has fallen from $220 per megabyte to nine cents per megabyte in the past 12 years. The price to store a megabyte of data on an optical disk will reach one cent with digital versatile disks that are the same size and appearance of CD-ROM disks, but store more than 10 times as much data, about eight gigabytes (eight billion bytes). The cost for a three-minute long-distance telephone call from New York to London has declined from $250 in 1930 to 15 cents today. With the deregulation of the telecommunication and cable TV industries, and widespread replacement of copper cable with fiberoptic cable that carries far more data at far lower costs, the price of communication will fall even more dramatically in the next few years.

Improvements in information systems lag

We would expect information systems vendors for health care organizations to take advantage of the increasing efficiency of microcomputer data processing, and improve their software as fast as they can. Instead, vendors are selling old, inflexible systems because they do not have the capital to adapt them to newer technologies, and they have substantial numbers of clients whom they are maintaining on their existing systems with considerable profit. Vendors take far longer to incorporate more modern technologies than we would wish. For instance, most of the vendors of patient accounting systems have not incorporated any of the following technologies: powerful microprocessors and parallel processing systems, relational database management systems, and networks that adhere to the standards of open systems.

If they use personal computers for client machines, they are only window dressing for emulating dumb terminals. Most of the processing is still performed on the server. They do not incorporate relational database management systems that make modifying systems and extracting data more efficient--they maintain their rigid hierarchical databases designed for the first version of their software. They maintain proprietary networks, rather than adopting open systems.

Vendors design their systems for one collection of hardware and software, and then they do not have the capital or incentive to rewrite their systems for better architectures that may come afterward. Microsoft, with billions of dollars in the bank for research and development, is trying to rewrite its client software now that network computing on the World Wide Web has become so important. Microsoft probably has the wherewithal to make the transition. Few other organizations developing information systems, and none serving the health care industry, have the capital to modify their existing systems so quickly. The incremental improvements they make to their products simply do not keep up with the rapid changes in the hardware and networking standards with which they must try to keep pace.

What do health care leaders need to do about these trends?

As a leader of an organization, one who approves substantial budgets for information technology, you must ask yourself what new products and services these startling technological advances will create. How will they change health and medical care? Physician executives need to consider these subjects, because they must guide the spending on information technologies and how those technologies are used for the benefit of patients and the competitive advantage of their organizations.

In 1960, the cost to lease an IBM 7090 mainframe, with transistors instead of vacuum tubes in its memory and logic circuits, was about $20,000 per month. The cost of the personnel to manage that 7090 was about $2,000 per month. The hardware was 10 times more expensive than the personnel. Back then, we learned that computers were expensive and people were cheap. Unfortunately, more than 35 years later, we still tend to invest too little in information technology, and squander money in payroll costs for tasks that could be automated.

A powerful personal computer costs about $3,500 now, and is capable of processing speeds and storage capacity (RAM and hard disk memory) greater than the mainframes of 10 years ago, let alone the 7090 35 years ago. You can lease such a PC for about $200 per month. The person using that computer costs the organization at least $2,000 per month, and up to $10,000 per month. Senior executives who might use such a computer to expedite their work, if they knew how, cost their organizations much more per month. So, the personal computer now costs the organization one-tenth to one-fiftieth the expense of the person who uses it. This ratio is more than a complete reversal in relative cost to the organization of people and the data processing equipment they use. Now people are expensive and computers are cheap.

You must lead your organization--where payroll costs are such a large proportion of total operating costs--in identifying ways to improve the productivity of employees and affiliated physicians with practical, prudent, and imaginative deployment of computers, software, and networks. A helpful way of thinking about such issues is to imagine what processes in your organization you'd like to automate if automation was free. Of course, nothing is free, but the costs of digital data processing are falling faster than expectations for digital data processing are rising.

Our thinking habits, more than the technologies themselves, limit our ability to conceive of many functions that could be automated to the benefit of our patients, organizations, families, and ourselves. We assume we can't do what we have not seen done before. We need to look for solutions. If microprocessors are involved, then the likelihood of finding a solution is improving annually at a stunning rate. What we can imagine today we can afford next year, and will take for granted the year after that. Let your imagination run freer than you usually permit it and see what clinical or administrative functions you wish to automate, then look for products and services to perform those functions.

Some examples

For instance, not long ago we used the DOS prompt to launch an application (word processing, spreadsheet, database, patient accounting system). We could have only one application operating at a time. We typed our print commands in arcane terms at the C:> prompt. A mouse was only a rodent.

Now we enjoy graphical interfaces, and point and click with mechanical mice on intuitive icons and menus. Simultaneously, we can keep operating our scheduling, patient accounting, word processing, electronic mail systems, and solitaire, and flip between them at the blink of an eye. In addition to text, we can store images, video, graphics, and sounds in our applications, such as our presentations. We can receive faxes at the same time we print our presentations and search the Internet's World Wide Web. We can use speech recognition to enter our commands, dictate our radiology interpretations, and complete our letters.

None of those functions we take for granted now was available 10 years ago. The World Wide Web was known to only a few people two years ago. Now, we are beginning to understand that it will give us the standards we have heretofore lacked for multimedia document exchange and interactive data sharing between computers in health care. All these improvements have occurred because of the maturation of microprocessors and related technologies. Today, we can hear about a new computer-based patient record one minute, and the next be operating a demonstration version of it over the Web. We can search the entire Medline database for key words and phrases in one or more of 30 medical lexicons in seconds from our homes, our offices, in fact, from anywhere on earth that has a telephone network. So, what's next?

What's next?

Can you imagine major improvements in a number of areas of medical management and clinical practice from inexpensive computing and communication technologies? What would you think about the following options? Computer-based patient records with structured data entry using voice recognition; expert systems for predicting treatment outcomes of patients; telemedicine sessions between clinicians from their workstations; home treatment counseling of patients and health plan members with interactive, multimedia, Web-based programs maintained on your practices' Web servers; continuous monitoring of patients' vital signs, blood oxygen saturation, electrolytes, and serum glucose, and a myriad other physiological parameters over dedicated telephone lines via the Web; automated interpretations of EKGs, EEGs, pap smears, and radiology studies by personal computers.

How about three-dimensional representations of detailed human anatomy that students can analyze on their clinical workstations; three-dimensional simulations of colonoscopy and cardiac catheterization; three-dimensional renderings of the molecular structure of enzymes and their receptors; or calculating the explanatory statistics that will predict the treatment outcome for a patient based on pre-existing patient characteristics. All these functions, which are dependent on fast microprocessors for acceptable performance, exist today either in commercial products or informatics laboratories.

It may help to know that as computer performance improves, most of the enhancements are absorbed by the human-computer interface. In 10 years, we have moved from text-based interfaces, with which users had to type arcane commands at line prompts, to graphical interfaces such as Apple Macintosh, Microsoft Windows, and IBM OS/2. There are other graphical interfaces, including NeXT Step, from NeXT (now being sold to Apple for $400 million), and Solaris, from SUN Microsystems (that also created the Java programming language for the Internet).

The next step forward in user functionality will be reliable and affordable speaker-independent voice recognition. That technology is maturing rapidly, and operating well on desktop PCs under IBM OS/2 Warp 4.0 today. Technology to rival it will be incorporated into Microsoft Windows and Macintosh OS shortly. By the way, vendors of computer-based patient records for ambulatory care with whom we have spoken admit they are changing their client computer operating systems from Windows to OS/2 to take advantage of IBM's lead in voice recognition that is native to the OS/2 Warp 4.0 operating system.

What about video conferencing?

We participate in multiple telephone conversations every day, usually making more than 10 calls daily at work, fewer from home. There are more than 250,000,000 long distance telephone calls each day in the U.S., and many more local calls. I would prefer to see, as well as hear, those persons with whom I have a relationship, or want to establish one. For those calls to people with whom I will have no relationship, those to directory assistance, or to a theater for its schedule, I do not need to see the person with whom I speak. Presuming that a video call is more expensive than a voice call, I'd rather use voice only for those brief calls to people who remain anonymous to me. But I'd prefer video calls with people I know.

We have the inexpensive video cameras and multimedia personal computers that can present video now. Most new notebook computers with MMX Pentium microprocessors have the circuitry needed for MPEG (Motion Picture Experts Group) 2 coding and decoding of video and audio signals. In the next few years, most Americans in urban areas will be offered ADSL (asymmetric digital subscriber line) modems from telephone companies for about $50 per month, from which a workstation can receive data over standard telephone wires at three to five million bits per second, and send data at 500,000 bits per second (hence the name asymmetric), which is plenty of bandwidth for two-way video and audio teleconferencing.

To compete, the cable TV companies will offer modems to attach to their coaxial cable, providing up to 30 million bits per second downstream into our computers, and about 400,000 bits per second upstream from our computers, also adequate bandwidth for bi-directional video and audio teleconferencing. The cable companies will lease those modems for about $50 per month or less.

How will microprocessors affect health care?

Microprocessors of every sort are affecting medical care. Not only are they the essential components of personal computers and workstations that are increasingly appealing to use, they are embedded in the controls of infusion pumps, digital thermometers, remote pharmaceutical dispensers on patient care units, imaging equipment of all sorts, home monitoring equipment, anesthesia machines, ventilators, laboratory auto-analyzers, automatic pap smear analysis, risk-adjustment of patient outcomes, and radiation treatment planning systems.

Think of the tasks that are automated now, with primitive interfaces, and how cheaper personal computers with graphical interfaces and speech recognition could help to make them more useful. Think of all those information management tasks that could be automated with the right equipment and networks to benefit patients, health plan members, staff, and physicians. The faster microprocessors bring with them calculation speeds to support voice recognition, expert systems, high-resolution three-dimensional rendering of data, rapid conversion of analog audio and video signals to digital format, and other capabilities.

You will register and schedule patients with voice responses. You will contract with health plans for capitation payments, guided by expert systems. You will accept responsibility for health promotion and prospective medicine for health plan members assigned to your organization and influence their demand for health care services with an extended Intranet, running on a Web server they can access over regular telephone lines and unique identification numbers and passwords. You will take and record histories and physicals on hand-held, backlit personal computers with touch screens for entering data. You will order diagnostic studies, perform procedures, and evaluate the results of diagnostic studies on hospitalized patients from your portable computer in your home, or in your hotel wherever you have traveled.

You will teach patients how to prepare for a procedure, what to expect from an ailment, how to minimize their disability and recuperate most quickly, and how to prevent recurrences of injury or illness, with multimedia instructional programs they access on your Web server any time. You will prescribe medications and send prescriptions electronically from your office practice system. You will collect functional status data on patient outcomes, study clinical data on your processes of care, and compare your performance to that of other clinicians with your same specialty using data from NCQA, JCAHO, state and federal government agencies, and cooperative payers that are stored in a relational database management system.

You will peruse clinical publications and complete continuing education from the monitor on your multimedia portable computer. You will manage human and inanimate resources of your medical practice with a sophisticated office practice management system, to maximize operational efficiency. You will communicate and negotiate with payers and patients to obtain all the funds rightfully owed to the practice via electronic text and voice mail from your workstation on your office network. You will calculate and complete records due to regulatory agencies for certification and licensure, and to federal and state agencies for tax purposes, and send them electronically according to EDI (electronic data interchange) standards.

Some aspect of each of these tasks mentioned above will benefit from automation on a personal computer with a fast microprocessor. Many of them, especially the simulations, image, and database analyses, will use the additional capabilities of readily available parallel processing systems to full advantage. The Internet's World Wide Web--a worldwide phenomenon--is made possible in largest measure by a sufficient number of persons, counted in the tens of millions around the globe, with powerful personal computers running Microsoft Windows, Apple Macintosh, and other operating systems that will run Web browser software from Netscape, Microsoft, Spry, Sun, and other companies.

The common browsers allow standardized information exchange of multimedia documents. With the advent of the Java programming language, information exchange is becoming more interactive and engaging, and the Web may well evolve into the standard architecture for modern computer-to-computer data exchange. In the process, the multimedia personal computer will become as much a communications device as a calculator or typewriter, and we will begin to use the Web for research, education, and teleconferencing.

In the near future, ever-improving microprocessors, either alone or in parallel arrays inside personal computers, will make the multimedia interfaces of personal computers much more intuitive, engaging, and easier to use. The personal computer is becoming a communications device--just as a global multimedia network, the Internet, is evolving to receive it. The results will be functionality in administrative and clinical systems not considered possible before, and ways of examining, treating, and teaching patients and health plan members far more potent than those today. Leaders of health care organizations need to direct their organizations to anticipate improvements in workstations and networks, and pressure vendors to incorporate those technologies into their products as quickly as possible.

Marshall Ruffin, MD, MPH, MBA, FACPE, is President of The Informatics Institute in Bethesda, Maryland. He can be reached at 8120 Woodmont Avenue, Suite 500, Bethesda, Maryland, 20814. He can be reached by calling 800/844-0922, via fax at 301/986-4546, or by e-mail at [e-mail].