The principal quasi-academic output which I can claim in the last decade of the century has continued to be centred around the question of the history of science in the Irish context, and its interaction with the culture. It is not 'peer-reviewed' in the full sense, there being no recognised academic local journal dedicated to this theme. There is currently an interaction with a Finnish journal 'Science Studies' and this may lead to a creditable version of the Boyle Medal paper. In 1992 I contributed a couple of entries on matters technological to the Blackwell Companion to Irish Culture, edited by WJ MaCormack, and these I suppose count, though I had forgotten about them by the time they were actually in print. I have reproduced them below.

• The Boyle Medal of the RDS as a measure of scientific esteem;
• Scientific Culture and National Identity: a hypertext construction based on several works on this theme, circa 1995;
• Science in India (Science and Public Policy Feb 1997);
• The Practical Arts in Irish Culture (Causeway, Belfast, Sept 1993);
• Physics on the Fringe (Physics Today July 1992, unabridged version).
• A paper on Biotechnology and Sustainability, published in 2001 by Farm and Food (ed Con O'Rourke).

Companion to Irish Culture: Entry on 'Industrial Revolutions'

The first industrial revolution can be characterised by the use of continuous mechanical power to turn machinery; typically millstones, a hammer for forging, or spinning and weaving machinery. These processes were subsequently rendered independent of a local water-power source by the invention of the steam engine. The rivers in Ireland, primarily in Ulster(1) but also in the neighbourhood of Dublin, Cork and many Irish towns, were on a scale such as to enable them to be harnessed easily to give the few tens of horsepower required for an industrial mill. Ruins of 18th century water-mills are a common sight in all parts of the country. Windmills are rare; Tacumshane in Wexford is preserved and Blennerville near Tralee has been restored.

There was another process at work which enhanced the industrial revolution(2) in Britain. The use of charcoal to smelt iron had stripped the forests; when in the 1750s smelting with coal was developed, rapid expansion became possible; the need to pump the mines gave rise to the steam engine. This process however largely bypassed Ireland, though attempts were made in the 1840s to mate the coal and iron which surround Lough Allen, and some effort was put by Thomas Mulvany(3) into the Shannon Navigation, with this and other industrial developments in mind. Mulvany went on to engineer the canals of the Ruhr and he is commemorated in Essen with a statue. Steam, based on native coal from Coalisland as well as imported, became the main motive power for the Ulster mills, although water remained important, the water-turbine being an Ulster invention (4).

A further enhancer of the Industrial Revolution in England was the accompanying agricultural revolution, associated with the name of 'turnip' Townsend. Feeding cattle in the winter with harvested fodder provided a steady supply of meat and milk to feed the expanding cities. The Irish livestock industry has to this day managed to avoid taking this modest step into the 19th century, with the result that the industry remains bedevilled by quality and continuity problems. The 'mild winters' are invoked by some experts as an excuse, but less credibly as added value and continuity become increasingly important to the market.

The main impact on Ireland of steam was the railways. However it could be argued that their role was to bleed the countryside and feed a parasitic metropolis. Consider their layout: the junctions outside Dublin are either in the middle of nowhere (Ballybrophy, Limerick Junction, Manulla, Farranfore) or else at the military garrison towns (Mallow, Athenry, Claremorris). The railway system never serviced direct interaction between the main urban centres outside Dublin. It serviced the emigration process and English-oriented commerce (5)

The electrical power revolution, when it came, had a very positive impact. The first wave was on the Edison direct current principle, and before 1914 most large and medium and some small towns were serviced, often with a small local hydro scheme on an 18th century mill site. The use of electric traction for a tramway was pioneered by the Traill brothers at Portrush in the 1890s(6); the power was hydro-generated, at Bushmills. There was technical support from Lord Kelvin and William Siemens for this important industrial pilot-project.

The 'second wave' (the 'Shannon Scheme') when it came in the 1920s for its time was quite visionary: it projected the production of several times the power then in use in all the urban generators, and laid down the basis for an 'national grid', taking advantage of the possibility of long-distance transmission at high voltage using alternating current (this was not completed in Britain until after World War 2). The approach to Siemens-Schuckert in Germany was an assertion of independence from the hitherto well-trodden channels to Britain. The prime mover was Dr Thomas A McLaughlan(7), a physicist turned engineer, who had served his time with Siemens in Pomerania, where a regional grid existed on the Irish scale.

Developments were interrupted by the war, but in the 1950s the distribution system to rural areas was commenced, and by the end of the 60s it was virtually complete. The use of single-phase 10kV lines was innovative, in response to the need to service isolated farmhouses(8). Elsewhere in Europe the rural community tends to live in villages, and isolated farmhouses are the exception. The case could be made that this was a 'technical fix' for a peculiar Irish social situation, arising from the over-dependence on a livestock-based economy. Had rural life been more village-based, there would have been more of an opportunity for small local industry to develop. Rural electrification has allowed the isolated farmhouse to remain marginally viable longer perhaps than it deserved. This argument is a hypothesis put forward by the present writer; it needs to be researched.

The communications revolution brought about by the telegraph, telephone and radio where it touched Ireland has been episodic: the transatlantic cable and the early Marconi enterprises were implants servicing the needs of Britain. The use of radio in 1916 in broadcast (rather than point to point) mode for the first time in history is said to have ensured that the news of the Rising got to the USA via shipping; the fact that the effort was made with equipment commandeered from the naval radio school in O'Connell St is well attested(9)

Less praiseworthy is the failure to make use of short-wave communications to keep the emigrants in touch. This became technically feasible in the 1930s, and was adopted effectively by the BBC for enabling the far-flung Empire to be kept informed of British politics and culture. An attempt was made by de Valera to set this up during the war, with the aid of E T S Walton in Trinity College, and equipment was purchased in the 50s for an RE World Service from Athlone, but later sold off at a loss(9).

The problem of supporting adequately the cultural needs of emigrants and expatriates remains with us, although the technology has been available since the 1930s.

The current wave of the industrial revolution is dominated by information technology. The first computer in the Republic, and perhaps in Ireland, a HEC, was in use in the Irish Sugar Co in Thurles in 1958. The Aer Lingus 'real-time' project of 1963-68, with which the present writer was associated, was a qualitative leap forward at the world state-of-the-art. The universities followed suit. There are currently (1992) some 300 firms in the Kompass directory listing 'software' among their products. Unlike earlier 'industrial revolutions', the information technology revolution is more than an implant; it seems to have put down roots in Ireland, and has the potential for becoming an accepted part of the culture.

Notes and References

Companion to Irish Culture: Entry on 'Technology'

As the technological processes became more industrial, and with greater dependence on scientific knowledge, the gap between them and what was perceived as Irish culture widened. This process can be followed by studying the interactions between the native Irish culture and the successive waves of foreign influences.

It can be argued that the fertility of Ireland, and the ease with which a livestock-owning culture could maintain itself in prosperity, removed the stimulus for taking an interest in those areas of technology associated with capital accumulation. A pure livestock culture, given mild winters, can get by with relatively simple technology.

Once crops become important, the development of technology to the proto-industrial level becomes inevitable: milling, iron ploughs, roads, bridges, brewing, distilling. Norman Ireland was moving in this direction (1), and would perhaps have evolved on the European model had it not been for the wars of Elizabeth and then Cromwell.

The process resumed again after Cromwell, but with a wider culture-gap between the planted ascendancy and the natives, comparable to what we have seen more recently in Kenya or Rhodesia (now Zimbabwe). In this situation, the control of technology was in the hands of the ascendancy; the native culture saw it as pure destruction. The forests were stripped to smelt iron and to build the English navy.

This culture-gap again began to narrow; as the 18th century progressed and manufacturing became established, primarily in the northeast but also elsewhere, primarily where there were good maritime communications (eg Cork). There emerged by the 1790s a technically competent bourgeoisie, which looked politically and philosophically to France, and sought to develop, on a secular democratic basis, a modern nation which was inclusive of the old native culture. This, as a colonial nationalist model, was innovative, and in contrast to the American practice which tended towards genocide.

The Royal Dublin Society at this time constituted a focus of applied research support for industry, of European stature, and was verging on the role of a college of technology; there was noticeable French influence (2).

This process of integration of technology into the culture was however again aborted by the unsuccessful attempt to establish a Republic on the French model in 1798, and by the subsequent Act of Union.

An important aspect of technology underpinning the emergence of modern European nation-states was the mastery of travel by sea to all parts of the world. In the post-Union diaspora, many talented people were exported(3) who contributed to the development of the maritime interests of the USA, Latin America, and various European maritime powers, as well as Britain. Had the political climate been congenial, these people would have serviced Irish needs.

In the 19th century, the emerging technological elite continued to be creamed off by the increasing pull of opportunity within the British Empire, in the USA, and on the continent.

Typical perhaps was the evolution of Charles Parsons, of steam turbine fame (4). A younger son of the Earl of Rosse, he served his time in the workshops associated with the great telescope at Birr Castle, while studying for engineering in Trinity College; but went into business in England. The Grubb optical works in Rathmines, also a spin-off from the Birr telescope, supplied the British Navy with gunsights, as well as most of the world with astronomical telescopes. This was 'cutting-edge' high technology, right up to the time it moved to St Albans 'for strategic reasons' in 1921.

The emerging Irish leadership in the 1916-1922 period had few, if any, people of influence who were in a position to appreciate the importance of industrial technology, increasingly based on science, as the key to national economic development,. Those who were technologically aware would have emerged primarily via Trinity College(5), and tended to look to the British Empire.

We can attribute this cultural gap to the ban on the 'Godless Colleges' imposed by Cardinal Cullen in the 1840s. This put barriers in the way of access to leading-edge scientific technology for the emerging Catholic component of the bourgeoisie. The Mechanics Institutes, the Dublin Technical Colleges, the Royal University, the College of Science and then finally the NUI emerged to fill the gap, but the 'intellectual partition' of the country remained a palpable barrier, right up to the 1950s.

The full integration of scientific technology into the perceived Irish cultural canon remains on the agenda(6). Sources useful in the quest for support for this process include the RIA(7) and the IEI(8).

Notes and References

Case-based Induction-tree Analysis in Diagnostic Protocol Development

This paper was contributed to the December 2000 'Expert Systems' conference, of the British Computer Society, in Cambridge.

Abstract

Introduction

Background

A. Rheumatoid Arthritis (RA)
B. Osteo Arthritis (OA)
C. Gout
D. Fibromyalgia (FM)
E. Polymyalgia Rheumatica (PMR)
F. Other Inflammatory Arthritis (IA)
G. Ankylosing Spondulitis (AS)
H. Cervical/lumbar spondylosis (C&L)
I. Local or regional non-articular pain syndrome (L&R)

The foregoing conditions were identified by the June-September 1999 version of the diagnostic protocol at a somewhat unsatisfactory 'hit rate', as measured against subsequent expert consultant diagnosis. Of the 45 cases, only 13 were 'hits' , with 6 'maybes' and 26 wrong.

We summarise this as follows:

Symptoms and Signs: 10 values
Age at onset: 4 values
Presentation: 3 values
Time Course: 4 values
Pattern of Joint Involvement: 8 values
Other features: 8 values (this is a list of additional possible signs and symptoms)
Investigations: 8 values (this is basically a classification of the results of lab tests)
Treatment effects: 4 values (responses to various trial treatments)
Other history: 4 values reflecting family histories of named conditions
Past history: 2 values reflection patient history of specific conditions

Each value of each parameter was allocated a weighting which was condition-specific, estimated on the basis of expert experience. For example the symptom 'single swollen joint' had a high weighting in favour of conditions B and C, and a high contra-weighting for conditions D, E and H

Those of us in the NHG Trust Group became aware that in IMS they had been getting some diagnostic experience in non-medical domains using an induction engine which had been developed in the context of the EU-funded INRECA project(1), the objective of which was to develop some tools originating in the 'artificial intelligence' research domain for use in a commercial environment. We felt that the utility of these tools might be worth exploring.

Inductive Analysis of Case-base

According to the induction-tree analysis, the first factor determining the outcome is 'pattern'. The largest group contained 22 cases having the value 'small + large joints'. In this group, the next most important determinant was 'time-course'.

There were 17 cases for whom 'time-course = 'continuous'. In this group, if there is no 'pain', there is in all cases a positive 'esr', of a wide range of values, leading to an RA diagnosis, except in one case, where the 'esr' is high at 84, and there is no 'stiffness'; here there is a possibility it could be RA or IA. (In the later group in fact we combine these two conditions, given that they are closely related).

There are 4 cases for whom 'time-course' is 'intermittent'; for these if the onset is sudden or rapid we have IA, and RA if gradual.

The one remaining case has timecourse = s12-14 and this is diagnosed as gout.

The next largest 'pattern' group is '2-4lj' containing 9 cases. 'Stiffness' comes next. Of these 7 have no stiffness; these are diagnosed as AS if 'pain' is 'none' or 'neckback', and as IA if 'pain' is 'general'. If 'stiffness' is 'local' we get LR, and if 'stiffness' is 'general' we get IA.

There are also 9 cases in the 'pattern' = 'diffuse' group. This group next splits by age. If age <40 there are 5 cases diagnosed as AS.

There are 2 cases in the group 40-60; one with 'pain' = 'neckback' is AS, the other with 'pain' = 'general' is RA.

There are 2 in the group >60; .joint' = 'none' gives FM and 'joint' = 'multiple' gives PMR.

Next we have 6 cases in the 'pattern' = '> 5 small joints' group. For these we look next at the 'timecourse'; one 'intermittent' is gout, two 'sl<48' are RA; there are 3 for whom 'timecourse' = continuous; of these 2 are RA if there is no 'stiffness', and maybe IA or RA if there is local stiffness.

Then finally there are 2 cases for whom the 'pattern' = '1 large joint'; for these if joint = none we have LR, and if joint = single we have OA.

This was a crude first pass, and it showed up quite clearly that the first factor to go for was the 'pattern', and that diagnoses were usually feasible using a relatively small set of indicators, with the other factors remaining in the background, for use in confirmation, for reassurance of the consultant that the initial diagnosis was correct.

Revision of Protocol

Pain (4 values)
Swelling (4 values)
Stiffness (5 values)
Pins and Needles, Tingling, Numbness (PNTN) (3 values)

(The foregoing are a subdivision of the earlier parameter 'signs and symptoms')

Joint Pattern (9 distinct values, increased from 8)
Tender points (3 values)
Muscle (2 values)

Each value of each parameter is associated with a set of weighting factors reflecting its significance in the context of each of the specified diagnostic outcomes.

Some re-grouping and re-naming of the diagnosed conditions was also done; this was a pragmatic act, for convenience, done without the knowledge that any scientific 'before and after' study was in prospect. With hindsight it would perhaps have been better to have stayed with the original naming and grouping of conditions.

There is however a rough correspondence, as follows:

RA and IA are grouped as IA (inflammatory arthritis)
Gout becomes CA (Crystal Arthritis)
AS (Ankylosing Spondulitis) remains the same.
PMR becomes CTD (*connective tissue disease)
OA is re-named OST (osteoarthritis)
C&L becomes C/LS (cervical/lumbar spoldylosis)
FM (fibropmyalgia) and L&R (local or Regional non-0articular pain syndrome) are grouped as NAR (non-articular rheumatism)
and a new condition N/R is identified (neuropathy/radiculopathy).

The foregoing revised diagnostic protocol was constructed using the insights derived from the above inductive analysis by the medical experts in the NHD Trust.

Performance of revised protocol

In order to measure the effectiveness of the first iteration of the inductive procedure, we took the cases which had been 'misses' under the old protocol, and ran them through the new one. The results for the first five were as follows:

A further group of 16 of old-protocol 'misses' was re-run under the new protocol. We were unable to correlate the re-runs case by case as above, but the overall picture is as follows:

This result suggests that in the next iteration of the induction engine, with the new protocol, we should look closely at the branching of the tree leading to AS, OST and IA, and consider giving higher weightings to the secondary and tertiary diagnostic factors which pick these out. At the time of writing we have not yet done this, but we hope to be able to report on it verbally.

The main difference between the old and the new protocols was the attention given to the details of the values of the parameters, based on the assessment of their importance, as derived from the first run of the induction engine.

There were also indications from the inductive analysis which suggested how the weighting factors should be revised. This processing of the output of the inductive analysis was subject to selective assessment using consultant experience. In other words, the role of the induction engine was as a tool in support of the process of expert adjustment of the structure of a diagnostic protocol, and of the weighting factors assumed within it.

The induction tree is in no sense a 'substitute' for expert diagnostic procedures; rather it is a support system for the structured quantitative abstraction of expert experience.

It is to be expected that as new cases arise, even though the weightings may have been adapted to give a perfect result for the cases to date, as the size of the sample population increases more 'misses' will again begin to appear. When these accumulate to some defined level, there will be a re-run of the induction engine, generating a further revision initially of the weightings, then if necessary of the protocol structure. By repeating the process successively, we in effect formalise an 'organisational learning by experience' procedure.

Conclusions

Further work is necessary, primarily in the elucidation of explicit procedures for translating induction-tree information into diagnostic protocol structures and weightings. There is also a need to extend this procedure into other diagnostic domains, and the elaboration of rules for recognising how such domains should be delimited, in such a way as ensure that the inductive tree serves as a clear indicator. Any attempt to make the domain too large would result in a decision tree of such complexity that the elucidation of valid weighting rules would be difficult.

There is also scope for elucidating weighting factors for allocation to parameters, as distinct from weighting factors for condition-identification to values within each parameter.

The definition of what constitutes a 'parameter' and what constitutes a 'value within a parameter' is another grey area, currently a matter for negotiation between the 'induction tree specialist' and the 'medical diagnostic specialist'. We do not yet pretend to have answers to any of these questions. We have simply reported that we have found the inductive analysis useful in improving diagnostic referral protocols in one specific delimited domain, and that as a consequence the consultants' work-load has been reduced, and the reliability of GP diagnosis improved.

Reference

Acknowledgement: the authors wish to thank Angela Millett of the Pontefract Trust who has been extremely helpful with the abstraction of case-base data into a form which made this paper possible at relatively short notice.

Some navigational notes:

Copyright Dr Roy Johnston 2002