Discrimination and diagnosis
Başlık:
Discrimination and diagnosis
Yazar:
O Muircheartaigh, Iognaid Gearoid, author.
ISBN:
9780438058019
Yazar Ek Girişi:
Fiziksel Tanımlama:
1 electronic resource (132 pages)
Genel Not:
Source: Dissertation Abstracts International, Volume: 76-08C.
Advisors: John Aitchison.
Özet:
The general problems discussed and developed in the course of this thesis are those of the application of statistical techniques to medical diagnosis. The situation with which we shall be mostly concerned will be that where we are given, for each of k diseases, (k ≥ 2), a random sample of patients together with a case-history (i.e. a p-variate symptom vector with p ≥ 1) for each of these patients. These case-histories will consist of observations, hopefully relevant to the diagnosis problem, on the patients' state of health and may be either quantitative (continuous) variables - viz. blood pressure, temperature, age, etc. - or qualitative (discrete) variables - viz. presence or absence of headache, severe, mild or no discomfort etc. We shall hope to discover effective rules for allocating a new patient to that one of the k diseases from which he is suffering. In Chapter 2 the linear discriminant technique is described, and a refinement of this method is suggested whereby we find the best linear discriminant, whether or not the covariance matrices of the underlyng (assumed multivariate normal) populations are equal. The possible connections between linear discriminant analysis and a certain aspect of the treatment allocation problem are examined in Chapter 3. Situations in which a linear discriminant would not be effective are examined in Chapter A new approach - which we call quadric analysis - is suggested as an improvement. The quadric analysis technique is developed and applied successfully to the preoperative diagnosis of the presence or absence of an adrenocortical tumour in patients suffering from primary hyperaldosteronism. In Chapter 5 we show a method for discovering what degree of discrimination can be achieved using subsets of the full variable set. All subsets are inspected by use of an efficient algorithm, and the best subsets of each size are found. The expected information divergence criterion is suggested as another method of discovering subsets of discriminating effectiveness. These techniques are applied to the data of section 6, showing that diagnosis is possible using as few as 4 of the original 8 variables. In Chapter 6 we give a brief account of two methods which might be used for a sequential diagnostic scheme using discrete variables. Appendices I, II, III contain the computer programs relevant to the analyses of Chapters 4 and 5.
Notlar:
School code: 0547
Tüzel Kişi Ek Girişi:
Mevcut:*
Yer Numarası | Demirbaş Numarası | Shelf Location | Lokasyon / Statüsü / İade Tarihi |
---|---|---|---|
XX(684651.1) | 684651-1001 | Proquest E-Tez Koleksiyonu | Arıyor... |
On Order
Liste seç
Bunu varsayılan liste yap.
Öğeler başarıyla eklendi
Öğeler eklenirken hata oldu. Lütfen tekrar deneyiniz.
:
Select An Item
Data usage warning: You will receive one text message for each title you selected.
Standard text messaging rates apply.