Tips on buying microcomputer hardware and software.
So many manufacturers are clamoring for your attention, and so much is changing. What was cost-effective in one way two years ago may no longer be cost-effective in the same way today. I will try to act as a guide here and provide general rules toward selection of the most appropriate hardware and software for the clinical laboratory.
Microcomputer technology is inexpensive and sophisticated enough for most data processing applications. Perhaps the only exception warranting a bigger computer is data base management in a large laboratory. But the majority of facilities can rely on a system that is simple to implement and suitable for a broad spectrum of laboratory management functions. This system may consist of one microcomputer with a hard disk for extensive data storage and several portable microcomputers to be moved around as the need arises. The software should include a word processor, spreadsheet, graphics, data base management, and communications to transfer data.
There are currently three principal types of microcomputers, classified by the power of their microprocessor: 8/8 bit, 16/8 bit, and 32/16 bit. I'll explain what that means. The microprocessor controls execution of all the operations of the microcomputer, such as data storage and retrieval, arithmetic, and logical instructions. It does so by manipulating a certain number of bits (units of information) at a time--usually 8, 16, 24, or 32. Many computers perform their internal functions by sending 16 bits at a time through a data path, while external operations (sending or receiving data) occur in groups of 8 bits. These are 16/8 bit computers.
In the 8/8 category are such microcomputers as the Apple II and Radio Shack models I through IV, which use the 6502 and Z80 microprocessors, respectively.
In the 16/8 category, we find the IBM PC with the 8088 microprocessor, as an example.
In the 32/16 category is the Apple Macintosh with the 68000 microprocessor, again as an example.
I will recommend hardware on the basis of three principles:
1. Invest in well-known technology. Very new technology may produce technical complications more suitable to a research environment than to a clinical lab.
2. Invest in compatibility. All microcomputers in the laboratory should be able to share data and software, preferably on the same diskettes but at least through cable/modem.
3. Purchase the most cost-effective microcomputer, taking into account current prices and available discounts. Consider bundled software, which is sold as a package with the hardware, as well as the cost of the hardware alone. Many computer manufacturers buy the rights to distribute a particular software program with their hardware for a fixed price. If that price is $100,000 and projected sales are 20,000 units, the added cost for the bundled software is $5 per unit--it is essentially thrown in free with the microcomputer. That's a bargain if you would have bought the software anyway for your applications; if the software is of no use to you, shop for the best hardware price.
Microcomputers using the Z80 or the 6502 microprocessor, like the Apple II, have been limited in total memory available for programs and overall speed of program execution. Even though their newer versions are not as limited, it is now more cost-effective to buy 16/8 bit and 32/16 bit units.
For example, a 16/8 microcomputer compatible with the IBM PC, having two disk drives and 128K RAM (random access memory), can be obtained for $2,000. The large RAM capacity coupled with sufficient disk access speed will satisfy most laboratory needs. As we noted, a large lab that must have data on line for thousands of patients will require more computer power. But very large capacity hard disks--over 50 megabytes, selling for well under $10,000--and 32/16 microcomputers could handle the hardware needs of almost any laboratory in the U.S.
The 16/8s, comparable in price to the 8/8s, are more powerful and have a wide range of available software. True, the Apple II in the 8/8 category has had better hardware attachments and software for instrument interfacing, but new data acquisition software for IBM PC compatibles may soon erase the Apple's edge in this area.
Although 32/16 microcomputers have greater RAM capacity than 16/8s, and their program execution is at least twice as fast, they constitute a fragmented market. No clear leader has emerged.
The result is little compatibility. Software and hardware developers cannot supply programs or attachments for all models, and one risks owning a computer with limited applications. Since each manufacturer sells a relatively small number of units, production costs are spread over a small base, and retail prices are high.
So my hardware vote goes to 16/8 IBM PC compatibles that can share data and software on diskettes. Besides IBM models, these include microcomputers manufactured by Columbia Data Products, the Zenith 150 and 160 (not to be confused with the 100 series, which isn't fully compatible with the IBM PC), the Compaq from Compaq Computer Corp., and the Panasonic Sr. Partner. They can take on many software and hardware accessories for lab applications. Whichever computer you select, make sure it has parity check, a feature that minimizes errors in the lab.
Let's turn now to programs. In an earlier Computer Dialog column ("The Advantages of Buying Rather Than Writing Software," MLO, August 1984), I advocated adaptation of general purpose commercial software for most laboratory uses. The power of available general purpose software commonly increases with the power of microprocessors, making it less and less cost-effective to write your own programs for particular applications. Time saved by not having to develop software more than offsets the cost of purchasing a program.
I recommend BASIC (or PASCAL) programming only when it is almost impossible to use general purpose software to accomplish a desired end. Two major applications that may require special programming are computer-assisted instruction and interpretive reporting. Here are some key areas for general purpose software:
* Word processing. Various reports, manuals, and form letters are typical WP applications. Current programs are quite sophisticated, and almost any software priced above $150 is adequate for the lab. If you know what kinds of reports you want to produce, selection poses no real problems.
Consider a "merge" feature. It enables you to insert data from one file--for example, the patient's name, room number, test results, and comments or a diagnosis--into another file, which may be a form letter. If the data are in several files, however, they'll probably have to be merged by a data base management system prior to word processing.
Among other features to look for are programs that keep all manuscript in random access memory, which is much faster than those that require repeated disk access; the ability to move columns, useful for editing reports; and the capacity to call up often-abbreviated terms from a glossary file, in order to insert explanations in reports.
WP software has two aspects, editing and printing. Most programs accomplish the same editing tasks--the difference, often a big one, lies in how long it takes to perform a task. Even if a program lacks a feature, it is likely to provide an alternative, though at a cost in extra user time.
The printing side of WP determines the format for hard copies and particular printers to use. It's less flexible than the editing aspect. You usually get only the features advertised; to get a different printer, for example, you may need a different WP program.
* Spreadsheets. Spreadsheet applications span the full range of laboratory management activities--from workload recording to staffing, budgeting, forecasting, and accounting--and also extend to such technical areas as quality control. Cost accounting procedures that would take weeks to develop in BASIC can be implemented in less than one day by adapting a commercial spreadsheet.
Features to look for include the ability to produce graphs from the data, either through integration of graphics in the software or by using another program; a programming language that allows the user to set up comparisons and analyses, using one or more tables; ability to refer to column or row headings by name instead of number (hematology instead of column 5, for example), which facilitates editing and modification of table formulas; ability to handle three-dimensional tables (pulling out data from tables under a new heading, such as all occurrences during a specific time frame); and integration and recalculation of data from multiple tables.
My earlier Computer Dialog column outlined advantages of software integrating spreadsheet functions, data base management, word processing, and graphics. Several of these programs are quite good, and selecting one is not difficult. They may sell for a few hundred dollars more than a single-application spreadsheet, but are worth the extra cost--even if the capabilities of one or another component, especially the data base management, are not wholly adequate for a laboratory's needs.
* Data base management. With general purpose DBM software introduced for 16/8 microcomputers since 1983, laboratories can design data bases suitable for most data storage and retrieval applications. Specific purpose data base management software used to be more effective for a particular application than a general purpose DBM program adapted to the application. Main advantages of specially designed programs were speed of record storage and retrieval, unique input screens and reports, and ability to merge data from many files without redundant data entry. Today, many general purpose DBM programs can satisfy those needs.
In addition, because many more copies of the general purpose software are sold, they are likely to contain fewer bugs or errors. If the vendor doesn't correct the errors when customers report them, sales will decline. The general purpose approach also permits further applications with the same software and easier, quicker modifications.
There are two basic kinds of data base management software (actually, many programs won't fit neatly into just one category): file management systems and relational data bases. File management systems are equivalent to a drawer of 3 X 5 index cards. If each card (record) contains 12 items (fields), the computer may organize the data according to any of those 12 items--and only those items.
Relational data bases, which I recommend for purchase, are more versatile and easier to modify. They have the capabilities of file management systems plus the advantage of combining data from two or more files. One file, for example, might contain records of patients who received blood transfusions; another, the records of donors. A relational data base could bring together information in both files, linking patients with their blood donors.
Multiple-file applications of data base management include inventory, test reporting, billing, and anatomic pathology reports (using a WP program to write the reports). A possible single-file application is a registry of microscopic slides for case studies.
It's more difficult to find the right data base management software than it is to buy good word processing or spreadsheet software. Read published program evaluations and consult with users.
Becoming an informed consumer can spare you a lot of trouble later on as a computer user.
|Printer friendly Cite/link Email Feedback|
|Author:||Siguel, Eduardo N.|
|Publication:||Medical Laboratory Observer|
|Date:||Sep 1, 1984|
|Previous Article:||Why I took my home computer to work.|
|Next Article:||The diagnostic marketplace in 1990.|