liu.seSearch for publications in DiVA
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • oxford
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Medical data time series analyses using AI techniques
Linköping University, Department of Biomedical Engineering, Medical Informatics. Linköping University, The Institute of Technology.
2002 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]

In many areas of medicine, we collect serial measurements of different variables as they change over time. Changes in a time series is often reflective of a change in a patient's status and may be helpful in guiding the appropriate therapy. As we tend to collect data more data as well as more frequently and for longer periods of time, the amount of information available to the clinician grows. Even though there is and has been a siguificant development in medical technology, the management of these growing anaounts of data still requires manual interpretation in many cases.

Typically, the information collected from serial measurements of data is used to establish a trend, which may be related to a patient's progress or deterioration. Isolated measurements of a variable also contiibute to a better understanding of a patient's status. The physician ahnost always manually interprets this type of infonnation. Time series may contain other types of information that may not be as easily be interpreted this way.

This thesis describes a number of cases where other types of information have been extracted from medical data time series. Time series may contain information that can be used for prediction as well as categorization (diagnosis). Except from the potential of improving patient care, this information is also helpful in expanding the understanding of the processes involved in each example.

The methods used in this thesis include principal component analysis, neural networks, cross-correlation analysis and the wavelet transform. By applying these methods to analyse the information in the various time series, adaptable and effective tools for an increased understanding of these problems may be created. This may contribute to better patient management by providing the clinician with improved decision support.

Place, publisher, year, edition, pages
Linköping: Linköpings universitet , 2002. , 49 p.
Series
Linköping University Medical Dissertations, ISSN 0345-0082 ; 726
National Category
Medical and Health Sciences
Identifiers
URN: urn:nbn:se:liu:diva-29437Local ID: 14783ISBN: 91-7373-164-1 (print)OAI: oai:DiVA.org:liu-29437DiVA: diva2:250252
Public defence
2002-04-15, Högdahlsalen, Universitetssjukhuset, Linköping, 13:00 (Swedish)
Opponent
Available from: 2009-10-09 Created: 2009-10-09 Last updated: 2012-09-14Bibliographically approved
List of papers
1. Categorization of fetal heart rate patterns using neural networks
Open this publication in new window or tab >>Categorization of fetal heart rate patterns using neural networks
2001 (English)In: Journal of medical systems, ISSN 0148-5598, E-ISSN 1573-689X, Vol. 25, no 4, 269-276 p.Article in journal (Refereed) Published
Abstract [en]

Digitized data from CTG (cardiotocography) measurements (fetal heart rate and uterine contractions) have been used for categorization of typical heart rate patterns before and during delivery. Short time series of CTG data, about 7 min duration, have been used in the categorization process. In the first part of the study, selected CTG data corresponding to 10 typical cases were used for purely auto associative unsupervised training of a Self-Organizing Map Neural Network (SOM). The network may then be used for objective categorization of CTG patterns through the map coordinates produced by the network. The SOM coordinates were then compared. In the second part of the study, a hybrid neural network consisting of a SOM network and a Back-Propagation network (BP) was trained with data corresponding to a number of basic heart rate patterns as described by eight manually selected indices. Test data (different than the training data) were then used to check the performance of the network. The present study shows that the categorization process, in which neural networks were used, can be reliable and agree well with the manual categorization. Since the categorization by neural networks is very fast and does not involve human efforts, it may be useful in patient monitoring.

National Category
Medical and Health Sciences
Identifiers
urn:nbn:se:liu:diva-29553 (URN)10.1023/A:1010779205000 (DOI)14924 (Local ID)14924 (Archive number)14924 (OAI)
Available from: 2009-10-09 Created: 2009-10-09 Last updated: 2012-09-14Bibliographically approved
2. Prediction of blood glucose levels in diabetic patients using a hybrid AI technique
Open this publication in new window or tab >>Prediction of blood glucose levels in diabetic patients using a hybrid AI technique
1999 (English)In: Computers and biomedical research, ISSN 0010-4809, E-ISSN 1090-2368, Vol. 32, no 2, 132-144 p.Article in journal (Refereed) Published
Abstract [en]

One of the problems in the management of the diabetic patient is to balance the dose of insulin without exactly knowing how the patient's blood glucose concentration will respond. Being able to predict the blood glucose level would simplify the management. This paper describes an attempt to predict blood glucose levels using a hybrid AI technique combining the principal component method and neural networks. With this approach, no complicated models or algorithms need be considered. The results obtained from this fairly simple model show a correlation coefficient of 0.76 between the observed and the predicted values during the first 15 days of prediction. By using this technique, all the factors affecting this patient's blood glucose level are considered, since they are integrated in the data collected during this time period. It must be emphasized that the present method results in an individual model, valid for that particular patient under a limited period of time. However, the method itself has general validity, since the blood glucose variations over time have similar properties in any diabetic patient.

National Category
Medical and Health Sciences
Identifiers
urn:nbn:se:liu:diva-29804 (URN)10.1006/cbmr.1998.1506 (DOI)15216 (Local ID)15216 (Archive number)15216 (OAI)
Available from: 2009-10-09 Created: 2009-10-09 Last updated: 2012-09-14Bibliographically approved
3. Analysis of the information content in sonoclot data and reconstruction of coagulation test variables
Open this publication in new window or tab >>Analysis of the information content in sonoclot data and reconstruction of coagulation test variables
2002 (English)In: Journal of medical systems, ISSN 0148-5598, E-ISSN 1573-689X, Vol. 26, no 1, 1-8 p.Article in journal (Refereed) Published
Abstract [en]

The Sonoclot Coagulation AnalyzerTM is a viscoelastometer used for in vitro analysis of the coagulation process from the start of fibrin formation, through polymerization of the fibrin monomer, platelet interaction, and eventually to clot retraction and lysis. In this paper, we have analyzed series of Sonoclot curves and simultaneously obtained coagulation tests (APT, PT, Fibrinogen, Platelet Count, and D-dimer) from patients who underwent total hip replacements (THA). By using the Principal Component Analysis method (PCA), we found that the most important coagulation test variables as reflected in the Sonoclot signature, are Platelet Count, PT, and Fibrinogen. Also, by using a Back-Propagation Neural Network (BP), we were able to reconstruct the coagulation variables Platelet Count, PT, and Fibrinogen from the Sonoclot curve with a reasonable accuracy. This would also indicate that these three coagulation test variables are most important in determining the appearance of the Sonoclot signature.

National Category
Medical and Health Sciences
Identifiers
urn:nbn:se:liu:diva-29552 (URN)10.1023/A:1013007202250 (DOI)14923 (Local ID)14923 (Archive number)14923 (OAI)
Available from: 2009-10-09 Created: 2009-10-09 Last updated: 2012-09-14Bibliographically approved
4. Analysis of the relationship between activity and pain in chronic and acute low back pain
Open this publication in new window or tab >>Analysis of the relationship between activity and pain in chronic and acute low back pain
(English)Manuscript (preprint) (Other academic)
Abstract [en]

We studied the temporal relationship between pain and activity in patients with acute or chronic low back pain. Our hypothesis was that activity exacerbates acute pain, but not chronic pain. To test this hypothesis, we concurrently measured activity and pain using continual electronic recording, and analyzed the data using the cross-correlation function.

After obtaining IRB approval and patient consent, we studied fifteen patients with acute low back pain and fifteen patients with chronic low back pain over 3 weeks. The activity levels were collected automatically using a wrist accelerometer, and were sampled every 1-minute. The pain levels were recorded semi-automatically using a computerized pocket-sized diary, every 90 minutes. The patients were prompted to enter a number between 0 and 10, where 0 is no pain, and 10 is the worst possible pain. Patients were allowed to enter additional measurements as often as they wanted.

The first seven and the last seven of the daily pain and activity time series from each patient were then analyzed using the cross-correlation function at various time lags between -60 and +60 minutes. The null hypothesis was tested by R to Z transformation, followed by a one-sample t-test against an expected Z score of zero.

We found that during the first seven measurement periods of acute low back pain, there was a significant (p<0.01) degree of cross-correlation between activity and pain. As these patients improved and reported less pain, the relationship between activity and pain disappeared. There was no such relationship at any point in time among the patients with chronic low back pain. These results confirmed our hypothesis in this small sample of patients.

National Category
Medical and Health Sciences
Identifiers
urn:nbn:se:liu:diva-81435 (URN)
Available from: 2012-09-14 Created: 2012-09-14 Last updated: 2012-09-14Bibliographically approved
5. Categorization and analysis of pain and activity in patients with low back pain using neural network technique
Open this publication in new window or tab >>Categorization and analysis of pain and activity in patients with low back pain using neural network technique
2002 (English)In: Journal of medical systems, ISSN 0148-5598, E-ISSN 1573-689X, Vol. 26, no 4, 337-347 p.Article in journal (Refereed) Published
Abstract [en]

Low back pain represents a significant medical problem, both in its prevalence and its cost to society. Most episodes of acute low back pain resolve without significant long-term functional impact. However, a minority of patients experience extended chronic pain and disability. In this paper, we have explored new techniques of patient assessment that may prospectively identify this minority of patients at risk of developing poor outcomes. We studied 15 patients with acute low back pain and 25 patients with chronic low back pain over 4 month's time. Patients monitored their pain and activity levels continuously over the first 3 weeks. Pain and functional status were assessed at baseline and at 3 weeks following enrollment. Follow-up assessment of functional status and progress were performed at 2 and 4 months. The pain and activity levels were categorized using a self-organizing-map neural network. A back-propagation neural network was trained with the categorization and outcome data. There was a good correlation between the true and predicted values for general health (r = 0.96, p < 0.01) and mental health (r = 0.80, p < 0.01). No significant correlation was found if activity and pain data were not entered into the analysis. Our results show that neural network techniques can be applied effectively to categorizing patients with acute and chronic low back pain. It is our hope that future research will allow these categorizations to be tied to prognostic and therapeutic decisions in patients who present with episodes of back pain.

National Category
Medical and Health Sciences
Identifiers
urn:nbn:se:liu:diva-29551 (URN)10.1023/A:1015820804859 (DOI)14922 (Local ID)14922 (Archive number)14922 (OAI)
Note

On the day of the defence day the status of this article was accepted.

Available from: 2009-10-09 Created: 2009-10-09 Last updated: 2012-09-14Bibliographically approved

Open Access in DiVA

No full text

By organisation
Medical InformaticsThe Institute of Technology
Medical and Health Sciences

Search outside of DiVA

GoogleGoogle Scholar

isbn
urn-nbn

Altmetric score

isbn
urn-nbn
Total: 523 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • oxford
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf