Printer Friendly

Trends in microcomputer-based lab systems.

Trends in microcomputer-based lab systems

Use of microcomputers in clinical laboratories is expanding rapidly. More than 50 per cent of laboratories surveyed by MLO had such units in 1983, compared with less than 15 per cent in 1980.(1) Extrapolation based on additional data from that survey suggests the current level is above 75 per cent.

There are many reasons for this strong growth in microcomputer popularity, including the continuously declining cost of hardware. As Werner noted,2 total expenditures for systems costing $10,000 or less have steadily increased over the last seven years. In 1986, these outlays may overtake what's spent on systems costing $30,000 or more.

Most popular microcomputers can be obtained for less than $5,000--often, at that price, with a color display screen, one removable and one fixed disk drive, a dot-matrix printer, and telecommunication capability or ability to communicate with other microcomputers.

Another major factor behind the rise of microcomputers in the laboratory is their widening capability. Powerful systems can be had for under $10,000.

We will review developments in computer technology that are bound to be of interest to clinical laboratories over the next few years. Our emphasis will be on laboratory information systems employing microcomputers rather than a larger minicomputer. Let's examine the state of the art in various system components.

Microprocessor chips. The key to increased power of future generations of microcomputers will be growth in the power of their central processing unit (CPU) chip. This chip, which is involved in all arithmetic and logic operations, is the central part of the computer system. Intel, Motorola, and National Semiconductor are fighting for leadership in the microprocessor market; all have more powerful chips that are ready or almost ready to go.

First-generation microcomputers used 8-bit chips, such as Intel's 8085, and could only process information in relatively small chunks. The number of bits, called CPU word size, made the chips slow by today's standards. The larger the word size, the faster the processing.

IBM's PC was the first widely sold microcomputer to use a 16-bit chip, Intel's 8088 (actually a hybrid design--half 8-bit, half 16-bit). The 8088 can handle information faster and address more memory than the 8085 chip, up to a 640-kilobyte user-memory limitation.

Intel's newer 80286 chip is driving IBM's PC AT, which was introduced last summer. The AT is just a hint of what's about to come in future PCs. Three times faster than the original PC, it relies mainly on the greater processing speed of the 80286 chip and the faster access time of its built-in hard disk.

IBM has come out with memory add-in products that give the AT as much as 3,000 kilobytes (3 megabytes) of random access memory. But the 80286 chip can actually use up to 16 megabytes of RAM--25 times as much as the original PC--and third-party vendors are reaching toward that limit.

A significant advance beyond any of these efforts is expected in 1986. Microcomputers built around Intel's new superchip, the full 32-bit 80386,3 will come on the market and realize the promise augured by the 80286-based AT. The 80386 is twice as fast as the 80286. Moreover, it can address 4 billion bytes (4 million kilobytes) of real memory.

The 80386 units will really be desktop minicomputers. Their power, in tandem with lots of memory, will redefine the role of the microcomputer.

Figure I shows some single-user microcomputers that already utilize Intel chips. All computer instructions must be decoded by the CPU, so fast internal clock rates are most desirable. On some microcomputers, such as the IBM PC AT and the Kaypro 286i, rates of over 6 million cycles per second can be achieved, which equates to a processing rate of 400,000 instructions per second. This means that even now, 80286-based microcomputers are approaching in speed the low end of the Digital Equipment Company's VAX minicomputer series. And as we have noted, the 80386 microcomputers will be twice as fast as that.

The minimum amount of main memory available on microcomputers has increased to 128 kilobytes, and most users are buying units with 256K or more. Some microcomputers, such as the Kaypro 286i, are designed for memory expansion up to 15 megabytes. That's as much memory as in some mainframe-computer hospital information systems.

Auxiliary storage. Most microcomputers have a removable diskette that also can serve for making backup copies. Improved manufacturing technology has increased diskette capacity to more than 1 million characters of information for a price below $300. This is useful but not enough by itself to meet the needs of a busy laboratory.

Fortunately, fixed hard-disk technology for microcomputer systems has also been developing at a fast rate. Just a few years ago, it was common to find only removable diskettes on small microcomputer systems. Today, it's desirable and justifiable to have one fixed disk and one removable disk. Desirable because the fixed disk can store 10 million or more characters of data and rapidly transfer these data into memory; justifiable because the cost has become quite reasonable from a price/performance standpoint.

With inclusion of a fixed disk, one begins to assemble a microcomputer system of sufficient power to be useful for a variety of laboratory tasks. However, experience with larger minicomputer systems in the laboratory demonstrates that even a fixed disk storing 10 million characters may very quickly fill up with data. When that happens, the data has to be cleared or off-loaded to an archiving tape to make room for new information on the disk.

You can acquire a good deal more capacity. Hard-disk drives storing as much as 960 million characters of data have become available for microcomputers, and that's sufficient to meet all the needs of a clinical laboratory. Companies producing large-capacity, IBM PC-compatible drives include Act Technologies, Alloy Computer Products, Corvus, Disc Tech One, Emerald Systems, Emulex, National Memory System, and Priam.

Prices in one sampling of their products ranged from $5,500 for a Disc Tech One drive with 135-megabyte storage capacity to $24,000 for National Memory's 960-megabyte drive. The latter had the lowest cost per megabyte --$25, compared with an average of $55. Of the external type, these drives are bigger than ones routinely used with a stand-alone microcomputer. Their abundant capacity is useful, and the cost can be justified if it is spread among several microcomputers on a network.

As for stand-alone microcomputers, they are increasingly being equipped with their own internal 10- to 40-megabyte hard-disk drives. The average cost is less than $50 per megabyte.

Display screens. Microcomputers with flat-panel displays have begun to appear on the market, and a flood of new designs should arrive over the next two years. These bright, highly readable electroluminescent and gas-plasma displays will replace dim and squinty liquid crystal screens; they will even surpass video display tubes in legibility. Manufacturers have resolved concerns about the price, fragility, and power requirements of the improved monitors.

IBM is marketing a large-screen plasma monitor that can display four full windows of data concurrently. Each window is 80 characters wide and 24 lines long.

Users are turning more and more to color and graphic options for better presentation of information and for computer-assisted instruction. The resolution continues to improve. A screen with a high resolution of 640 dots across by 400 dots down is now available with a choice of as many as 16 colors from a total palette of 256. At medium resolution, all 256 colors are available at once. There's a much broader spectrum in Commodore's new Amiga microcomputer: It has a palette of 4,096 colors.

Touch-sensitive screens are on the market, but most microcomputer-based systems use a "mouse' to point to display areas. Minimal training enables a user to easily wield a mouse and interact meaningfully with Apple's Macintosh system, for example. Although the Macintosh only displays in black and white, many gradations of shading are available, and the resolution is exquisite.

Voice recognition. Voice input units, designed to work with a circuit board that is easily inserted into a microcomputer chassis, may be acquired for relatively low cost. Vendors of IBM PC-compatible units include AudoPilot, Interstate Voice Products, Keytronic, Microphonics, Texas Instruments, Verbex, and Votan. Thanks to new chip design criteria, the price has dropped below $600 for products from AudoPilot and Microphonics. Each user must train the unit to accept his or her voice, but once that's done, a good level of accuracy can be achieved in word or phrase recognition.

Voice input units help out when a user's hands are too busy to enter data on a keyboard. An anatomic pathologist who is viewing biopsies might dictate observations to the microcomputer, for example, or a microbiology technologist might verbally record data about a culture.

Laser printers. Among the printers available are Apple's Laser Writer, Hewlett-Packard's Laser Jet, Personal Computer Products's Daisy Laser 1000, Quadram's Quad Laser, and Xerox's 4045 Laser CP. All have a resolution of 300 300 dots and an RS232C interface. Their speed is normally up to 10 pages a minute, but options can raise it up to 24. They operate silently, and they require little maintenance. At 2 feet cubed, the Hewlett-Packard printer is small enough to place on a desk or benchtop.

Prices range from $3,495 for the H-P and Personal Computer Products models to $6,995 for the Apple. Several microcomputers connected to a network can share a printer, reducing the cost per user.

System interfaces. So far, we have discussed the microcomputer as a stand-alone system. But we need the ability to communicate with other microcomputers and analytical instruments in the laboratory. Beyond that, one of the most important features of any laboratory computer system is its interdepartmental communication ability. These functions involve different kinds of interfaces.

Microcomputer to microcomputer. One way to interface microcomputers with each other is to connect them by cable in a local area network (LAN)--the local area might be within the laboratory, for example. Most of the microcomputers can operate somewhat on their own but will frequently act as terminals in this arrangement. They transmit and receive data over the network via a faster, more expensive microcomputer, which has access to an external disk and a printer. This principal microcomputer is called a resource server. Figure II lists some currently marketed LANs.

Another interfacing approach ties a number of users to one server or microcomputer. These users work at terminals, not microcomputers that can function independently of the central unit. Examples of such multi-user microcomputer systems are listed in Figure III. These tend to have larger CPU word sizes and more memory than single-user systems, and they are more expensive.

Multi-user microcomputers continue to be a viable product, despite predictions that LANs would ultimately ease them out. They are a better choice than LANs in any vertical application (within a laboratory section rather than across the lab) requiring computing power for a specific purpose, such s large-scale data entry or manipulation of large data bases.

Even in nonvertical applications, the easiest way to integrate single-user and multi-user microcomputers is to let the smaller units operate as terminals of the larger ones. This is inexpensive, efficient, and probably in keeping with the evolution of the technology. More sophisticated options exist, however.

For example, Altos Computer Systems recently introduced a super-microcomputer that can handle 30 users with its increased CPU power. Alternatively, the unit can become a file server and node, or work station, in a local area network. (A circuit board developed by the company fits into microcomputers, allowing them to act as LAN nodes.)

It's generally believed that LANs will one day be the best method for data and resource sharing. But they have thus far been plagued by high cost, high failure rates, and spotty performance.

A multiple-processor approach offers certain advantages over LANs. Here, each user has a microprocessor that interconnects with the other microprocessors. The link is via a standard "bus,' which is a conductor used to transmit signals between processors.

Alloy Computer Products is one company that markets equipment to form a multiprocessor system. When a card is inserted into a central microcomputer, dumb terminals can be attached by twisted-pair cable as "slave processors' capable of running the computer's programs.

This package is 10 times faster than a typical local area network. But despite the speed and other advantages of multiprocessor systems, networking will probably be the wave of the future. It can offer more resources than a single package can contain. Networks will have to be sophisticated, however, to meet laboratory needs.

The main problem in developing LANs for single-user microcomputers is how best to get different makes--Apples, IBMs, AT&Ts, and so on--to communicate with each other. Competition in this area is hot right now. That's due somewhat to Microsoft's recently released DOS 3.1 operating system, which incorporates connectors to hook on to communication software.

Vendors are scrambling to take advantage of this new product. Although Novell sells its own LAN, it has expressed support for DOS 3.1 as the industry-standard microcomputer operating system. It also recently announced a bridge product that lets users link one otherwise incompatible microcomputer network with another network.

Some LANs connect microcomputers having different types of operating systems (Figure II). This is an important consideration for a laboratory that has amassed a heterogeneous mixture of stand-alone microcomputers.

Microcomputer vendors are presently seeking a standard that would make a super-LAN a reality. Such a standard would enable users to plug any microcomputer into any network.

There are microcomputer-based file-serving units especially designed to link single-user microcomputers. One interesting unit, 3Server from 3Com, can support up to 1,024 microcomputer work stations.

A large number of work stations may be attached to LANs, at a cost varying from $14 to $215 per attachment, depending on the vendor and product. All of the LANs in Figure II except those marketed by IBM and 3Com use twisted-pair cable. Although twisted-pair cable is less expensive than coaxial cable, it usually results in a slower data transmission rate, except in the Proteon system.

Microcomputer to instrument. Analytical instruments also can be interfaced to microcomputers for smoother data acquisition. The simplest way is to connect a single instrument to a single microcomputer. This eases the tasks of acquiring, storing, and processing data, but there's a disadvantage: The microcomputer is dedicated to the data acquisition function and therefore cannot be used for other work at the same time.

A more complex arrangement interfaces the lab instrument to a network for data acquisition. If properly designed, this arrangement makes better use of a given microcomputer system. It may be more expensive and take more time to implement than a single instrument-microcomputer hookup, but when several instruments are involved, the network approach should be more economical per instrument interface.

A recent paper described an interface system for acquisition, temporary storage, and transfer of data from several instruments to a single microcomputer.4 Only on demand does this device transmit the data it has stored to the microcomputer for reduction and analysis. The microcomputer can thus be kept free, when the need arises, to perform other tasks or to analyze data from another instrument.

Microcomputer to mainframe. Perhaps the most important aspect of a lab computer system is interdepartmental communication. Figure IV shows how a network marketed by AST might be used to connect a lab microcomputer system to a mainframe hospital information system. Compatible microcomputers in various areas of the lab would be interconnected through a LAN to a resource server, which is then connected to the mainframe through a gateway microcomputer.

To accomplish this, a communication card must be installed in the gateway microcomputer. This, along with emulation software, establishes mainframe communication for microcomputers on the network. Since the gateway microcomputer is not dedicated, it remains available for other work as needed. The system also provides for concurrent microcomputer and mainframe sessions, through a "hot key' feature. Each microcomputer on the network can run its own programs in foreground mode and, in intervals between that activity, can communicate with the mainframe system in background mode.

In the next issue of MLO, we will review several commercially packaged microcomputer-based laboratory information systems.

1. Hallam, K. Microcomputers in the lab: The sudden boom. MLO 16(5): 30-36, May 1984.

2. Werner, M. Is it time to buy a computer network? MLO 16(9): 52-60, September 1984.

3. Kanzler, S., and Spector, G. Intel debuts the 80386 for simultaneous program execution. PC Week 2: 3, Oct. 22, 1985.

4. Danforth, D.R.; Stouffer, R.L.; and Bosnos, M. Laboratory instrument interface system (LIIS): A unit for the acquisition, temporary storage, and transfer of data to a microcomputer. Computers in Biology and Medicine 15(2):95-101, 1985.

Table: Figure I A sampling of single-user microcomputers

Table: Figure II Some local area networks

Table: Figure III A sampling of multi-user microcomputers

Table: Figure IV Microcomputer-to-mainframe interfacing
COPYRIGHT 1986 Nelson Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1986 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Groves, William E.
Publication:Medical Laboratory Observer
Date:Feb 1, 1986
Previous Article:Collaborative negotiation; getting agreements that last; supervisors and those they negotiate with on lab problems can reach an accord by focusing on...
Next Article:Managing a graying work force.

Related Articles
Microcomputers in the lab: the sudden boom.
Putting the magic to work.
Microcomputers in microbiology: a matter of special needs.
Workload recording with an electronic spreadsheet.
Streamlining blood counts with a microcomputer.
A look at packaged microcomputer systems.
Integrated software: a quick way to audit turnaround time or paid hours.
Quicker QC on a small microcomputer.
A microcomputer test-costing program.
Saving time with combined microcomputer applications.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters