Diagnosis of Alzheimer's disease is based on the results of neuropsychological tests and available supporting biomarkers such as the results of imaging studies. The results of the tests and the values of biomarkers are dependent on the nuisance features, such as age and gender. In order to improve diagnostic power, the effects of the nuisance features have to be removed from the data. In this paper, four types of interactions between classification features and nuisance features were identified. Three methods were tested to remove these interactions from the classification data. In stratified analysis, a homogeneous subgroup was generated from a training set. Data correction method utilized linear regression model to remove the effects of nuisance features from data. The third method was a combination of these two methods. The methods were tested using all the baseline data from the Alzheimer's Disease Neuroimaging Initiative database in two classification studies: classifying control subjects from Alzheimer's disease patients and discriminating stable and progressive mild cognitive impairment subjects. The results show that both stratified analysis and data correction are able to statistically significantly improve the classification accuracy of several neuropsychological tests and imaging biomarkers. The improvements were especially large for the classification of stable and progressive mild cognitive impairment subjects, where the best improvements observed were 6% units. The data correction method gave better results for imaging biomarkers, whereas stratified analysis worked well with the neuropsychological tests. In conclusion, the study shows that the excess variability caused by nuisance features should be removed from the data to improve the classification accuracy, and therefore, the reliability of diagnosis making.

Diagnosis of Alzheimer's disease is based on the results of neuropsychological tests and available supporting biomarkers such as the results of imaging studies. The results of the tests and the values of biomarkers are dependent on the nuisance features, such as age and gender. In order to improve diagnostic power, the effects of the nuisance features have to be removed from the data. In this paper, four types of interactions between classification features and nuisance features were identified. Three methods were tested to remove these interactions from the classification data. In stratified analysis, a homogeneous subgroup was generated from a training set. Data correction method utilized linear regression model to remove the effects of nuisance features from data. The third method was a combination of these two methods. The methods were tested using all the baseline data from the Alzheimer's Disease Neuroimaging Initiative database in two classification studies: classifying control subjects from Alzheimer's disease patients and discriminating stable and progressive mild cognitive impairment subjects. The results show that both stratified analysis and data correction are able to statistically significantly improve the classification accuracy of several neuropsychological tests and imaging biomarkers. The improvements were especially large for the classification of stable and progressive mild cognitive impairment subjects, where the best improvements observed were 6% units. The data correction method gave better results for imaging biomarkers, whereas stratified analysis worked well with the neuropsychological tests. In conclusion, the study shows that the excess variability caused by nuisance features should be removed from the data to improve the classification accuracy, and therefore, the reliability of diagnosis making.

Purpose

To compare the accuracies of predicting AD conversion by using a decision support system (PredictAD tool) and current research criteria of prodromal AD as identified by combinations of episodic memory impairment of hippocampal type and visual assessment of medial temporal lobe atrophy (MTA) on MRI and CSF biomarkers.

Methods

Altogether 391 MCI cases (158 AD converters) were selected from the ADNI cohort. All the cases had baseline cognitive tests, MRI and/or CSF levels of Aβ1-42 and Tau. Using baseline data, the status of MCI patients (AD or MCI) three years later was predicted using current diagnostic research guidelines and the PredictAD software tool designed for supporting clinical diagnostics. The data used were 1) clinical criteria for episodic memory loss of the hippocampal type, 2) visual MTA, 3) positive CSF markers, 4) their combinations, and 5) when the PredictAD tool was applied, automatically computed MRI measures were used instead of the visual MTA results. The accuracies of diagnosis were evaluated with the diagnosis made 3 years later.

Results

The PredictAD tool achieved the overall accuracy of 72% (sensitivity 73%, specificity 71%) in predicting the AD diagnosis. The corresponding number for a clinician's prediction with the assistance of the PredictAD tool was 71% (sensitivity 75%, specificity 68%). Diagnosis with the PredictAD tool was significantly better than diagnosis by biomarkers alone or the combinations of clinical diagnosis of hippocampal pattern for the memory loss and biomarkers (p≤0.037).

Conclusion

With the assistance of PredictAD tool, the clinician can predict AD conversion more accurately than the current diagnostic criteria.