Models of Spatial Diffusion have been developed and used in Geography at least from the time of the Swedish geographer Hiegastrand (I hope I have got his name correct), and the kindly old professor of Geography at Trondheim, Prof. Karlsen Asbjorn (now retired) who hosted me for several meals at his home when I was on sabbatical in Norway way back in the last century. Asbjourn was considered an expert in the field of applying those models to the diffusion of infections. There are several such models.
To use those models, one needs a strong grounding in mathematics which I don’t have. But something that Asbjourn told me – i.e. the need to make a series of assumptions what could invariably be no more than speculative – made me somewhat skeptical about the utility of such models, except by way of getting your publication in prestigious journals. It is from that viewpoint that I looked at the following extract from Magdon-Ismail’s paper
To use those models, one needs a strong grounding in mathematics which is something I don’t have. But something that my Norwegian professor told me — i.e. the need to make a series of assumptions what could be no more than speculative — makes me somewhat skeptical about the utility of such models, except by way of getting your publication in prestigious journals. It is from that viewpoint that I looked at the following extract from Magdon-Ismail’s paper.
“Professor Malik Magdon-Ismail has developed a models for several states in the badly-hit US that are sobering and alarming as was the case with New York with a population of 19 million – a few millions short of the Sri Lankan population. In the case of Sri Lanka his model shows the number of coronavirus cases in Sri Lanka peaking around mid-August. He hopes he is wrong but the model has its own outlook. This is determined by assumptions on how many people stay home and how quickly serious cases climb”.
Our population distribution is quite different from those of most states in the US federation where there are extraordinarily high densities in the cities that are surrounded by very extensive spaces with low densities. Moreover, there are significant behavioural contrasts between our people and those over there. What I understand from the Ismail model is that it is based on US data.
Moreover, the sentence“This is determined by assumptions on how many people stay home and how quickly serious cases climb” reads like a banality camouflaged by a pretext at sophisticated modelling.
OR, is it intended to provide ammunition to those in Sri Lanka who demand further postponements of the parliamentary elections (beyond mid-June) and/or re-summoning the dissolved parliament, among whom are Rauff Hakeem and the thoroughly discredited Rishard Baduideen, to mention only a few of the Yahapalana wallahs?
A NOTE: Via his remark that this style of modelling is about as helpful as the work of our Sri Lankan astrologers, Professor Chandre Dharmawardena has presented a reading similar to that of Gerry Peiris.
Thuppahi's Blog · This web site presents the interventions of MICHAEL ROBERTS in the public realm with reference to Sri Lankan political affairs. It will embrace the politics of cricket as well. ROBERTS was educated at St. Aloysius College in Galle and the universities of Peradeniya and Oxford. He taught History at Peradeniya University and Anthropology at Adelaide university. He is now retired and lives in Adelaide.
One response to “The Problem with Spatial Diffusion Models”
Pingback: Debating the Progress of Covid-19: Vibrant Viewpoints | Thuppahi's Blog