Background: Dementia is a priority for global public health. The management of behavioral and psychological symptoms of dementia (BPSD) is one of the highest ongoing challenges and needs new approaches. The special care unit for people with dementia and BPSD (SCU-B) is viewed in this context as a further medical intervention. Aim: this study aims to explore SCU-B units in order to describe their main characteristics in relation to different implementation contexts, identify the characteristics of their replicability, and examine the social innovation elements promoted by SCU-B units. Method: This qualitative study is based on focus groups (FGs) and interviews involving nine international centers. Five of the centers have a memory clinic unit and SCU-B, compared with six that only have a memory clinic unit. A total number of 18 FGs were held, which altogether involved 164 participants. All data were transcribed verbatim and analyzed by means of a content analysis and a SWOT (strengths, weaknesses, opportunities, and threats) analysis. Results: The qualitative analysis offers a vision of the SCU-B model as an innovative care unit for BPSD, promoting social innovation in the long-term care (LTC) sector. This system mainly targets people with dementia and BPSD and their informal caregivers but encourages collaboration between dementia care stakeholders at the micro and meso levels. Conclusions: Specific characteristics of the country’s LTC systems and the organization of specialized units are determinants for the success of the SCU-B experience. The replicability of the entire SCU-B model was considered low; however, the implementation of single elements composing the SCU-B model may foster innovation. This study provides relevant suggestions on how to implement the SCU-B unit and innovative solutions for dementia care.

Objectives The main goal of machine learning approaches to classify people into healthy, increased Alzheimer’s disease (AD) risk, and AD is the identification of valuable predictors for valid classification, prediction of conversion, and automatization of the process. While biomarkers from cerebrospinal fluid (CSF) are the best-established predictors for AD, other less invasive, easy-to-assess candidate predictors have been identified. Here, we evaluated the predictive value of such less invasive, predictors separately and in different combinations for classification of healthy controls (HC), subjective cognitive decline (SCD), mild cognitive impairment (MCI), and mild AD. Methods We evaluated the predictive value of personality scores, geriatric anxiety and depression scores, a resting-state functional magnetic resonance imaging (fMRI) marker (mPerAF), apoliprotein E (ApoE), and CSF markers (tTau, pTau181, Aβ42/40 ratio) separately and in different combinations in multi-class support vector machine classification. Participants (189 HC, 338 SCD, 132 MCI, 74 mild AD) were recruited from the multi-center DZNE-Longitudinal Cognitive Impairment and Dementia Study (DELCODE). Results HC were best predicted by a feature set comprised of personality, anxiety, and depression scores, while participants with AD were best predicted by a feature set containing CSF markers. Both feature sets had equally high overall decoding accuracy. However, all assessed feature sets performed relatively poorly in the classification of SCD and MCI. Conclusion Our results highlight that SCD and MCI are heterogeneous groups, pointing out the importance of optimizing their diagnosis criteria. Moreover, CSF biomarkers, personality, depression, and anxiety indicate complementary value for class prediction, which should be followed up on in future studies. Key Points Using multi-class support vector machine, we compared the predictive value of well-established versus non-invasive, easy-to-assess candidate variables for classification of participants with healthy cognition, subjective cognitive decline, mild cognitive impairment, and mild Alzheimer’s disease. Personality traits, geriatric anxiety and depression scores, resting-state mPerAF, ApoE genotype, and CSF markers were comparatively evaluated both separately and in different combinations. Predictive accuracy was similarly high for a combination of personality, anxiety and depression scores as for CSF markers. Both established as well as candidate variables performed poorly in classifying SCD and MCI, highlighting heterogenous causes of those cognitive states. CSF biomarkers and extended personality measures show complementary value for class prediction, which should be followed up on in future studies.