Literature review: State of current industry knowledge (Health Informatics)

profilekshlme
Alzheimersdisease_Holmes2.docx

More than 5.4 million people in the United States are living with Alzheimer’s which cannot be fully treated or reversed with the current medical and treatment system, however, with effective prevention technology its prevalence can be controlled. Various preventive or predictive models has been developed in the past however, almost all of them had several issues which made them less attractive (Wang, Qiu & Yu, 2018). In the study conducted by Wang, Qiu & Yu (2018) an RNN with LSTM cells-based progression model was proposed to predict the future stages of the Alzheimer’s disease. The model is found successful and authors showed their interest to use this model for the prediction of other chronic disorders.

Previously, machine learning algorithms and traditional time series methods were used in AD progression or predictive models (Wang, Qiu & Yu, 2018). Over the last fiver years, the research in dementia risk prediction and specially for the Alzheimer’s disease got a boost. Numerous old models have been improved while some new models were also developed to increase prevention of this disease by predicting its numerous stages. In 2010, a systematic review identified that there are more than 50 models developed for the prediction of dementia and these models exhibited a difference in terms of risk score calculation, model accuracy level, follow up time, and disease outcome (Nie et al., 2017). Although, there are various models developed by many researchers, however, still there is no such model about which it can be said that it is a perfect model for dementia/Alzheimer’s risk prediction among people due to the absence of risk score validation researches. There is a need to have lots of work to improve the existing models as well as to develop new models considering the weaknesses of existing models (Tang et al., 2015).

Researchers has found a strong connection between aging population and Alzheimer’s disease and it is assumed that innovation could help in the partial resolution of the aging related societal problems. A biomedical model of Alzheimer’s disease is discovered to reduce the social-culture aspects of the aging and Alzheimer’s disease. This model proposed a cure to Alzheimer’s disease; however, this model was abandoned due to the emergence of care models in 1990s and onward. Most models were discovered to make an early diagnostic of Alzheimer’s disease, however, due to the lack of any proper treatment the early diagnostics of this disease were remained useless as no cure was offered. The only way with which the early diagnostic or predictive technologies could help in the treatment of Alzheimer’s disease is that these early diagnostic models should eb a part of Alzheimer’s disease management in its early stages (Cuijpers & van Lente, 2015). Alzheimer’s disease is accountable for 70% of the dementia cases across the world therefore, the therapeutic paradigm of this disease has been moved to secondary prevention. Various social implications and ethical concerns has been identified in preclinical predictive approaches, therefore, there is a need of a proper predictive modeling is required to address the challenges brought be prevention approaches. Many predictive models have raised some social implications and ethical concerns that most of the studies on this matter has overlooked. The predictive models like Statistical algorithms method or the methods of source data such as imaging data, family data, genetic data, cerebrospinal fluid examination, neurocognitive assessment, electronic/medical health record, demographics and family history have raised some sort of ethical and social concerns (Angehrn et al., 2019).

The study of Tang et al. (2015) investigated 1,234 articles to investigate the new developments in risk prediction of dementia and Alzheimer’s disease and concluded that new developments like non-APOE genes testing, incorporation of diet, application of non-traditional risk factor of dementia, ethnicity and physical function, and inclusion of some development in subgroups with diabetic individuals and those with different educational levels. However, after making an investigation of various studies, it is concluded that there is no single study that could be used for the prediction of dementia. Moreover, it seems impossible to have a predictive model that fits all. There is a high need to make researches for the development of existing predictive models as well as for developing new models, however, it is important to consider various predictive models (Tang et al., 2015).

According to the study of Haas et al. (2016) discovered that many predictive models for Alzheimer’s disease have failed due to the absence of reliable models. This is because most models are computerized and there is a need for a more humanized kind of models that best suits the clinical situation. The biggest need is to make sure that data is shared among all the healthcare organizations because it allows researchers to address the various scientific questions in different predictive models. The availability of data will also enable the researchers to access both real life data and clinical trials data to innovate a more accurate and ethically sound predictive model. The diagnosis and prediction of Alzheimer’s disease is very difficult because of its different symptoms in different patients. Moreover, the progression of this disease from preclinical, mild and severe is different among different individuals. Instead on investigating appropriate predictive models, there is a need to invest in the training of physicians to make them ready and competent enough to manage Alzheimer’s disease at various stages. The improvement in the understanding level of the physicians could improve the overall cycle of treatment and management of the Alzheimer’s disease (Margolis, 2017).

Due to the lack of discovery of the pathogeneses of Alzheimer’s disease there is no proper cure available for it to date. Since it is widely believing that pathogenic changes began to occur years before therefore, it is important to have an accurate and effective predictive model to accurately screen, diagnose and prevent this disorder. Unfortunately, due to the polygenic nature of Alzheimer’s disease, it has not been possible to effectively predict the level of Alzheimer’s disease (Nazarian & Kulminski, 2018).

References

Angehrn, Z., Nordon, C., Turner, A., Gove, D., Karcher, H., & Keenan, A. et al. (2019). Ethical and social implications of using predictive modeling for Alzheimer’s disease prevention: a systematic literature review protocol. BMJ Open9(3), e026468. doi: 10.1136/bmjopen-2018-026468

Cuijpers, Y., & van Lente, H. (2015). Early diagnostics and Alzheimer's disease: Beyond ‘cure’ and ‘care’. Technological Forecasting and Social Change93, 54-67. doi: 10.1016/j.techfore.2014.03.006

Haas, M., Stephenson, D., Romero, K., Gordon, M., Zach, N., & Geerts, H. (2016). Big data to smart data in Alzheimer's disease: Real-world examples of advanced modeling and simulation. Alzheimer's & Dementia12(9), 1022-1030. doi: 10.1016/j.jalz.2016.05.005

Margolis, R. (2017). Exploring Outcomes and Value across the Spectrum of Alzheimer’s Disease. Presentation, 1201 Pennsylvania Avenue NW, Washington, DC 20004.

Nazarian, A., & Kulminski, A. (2018). POLYGENIC PREDICTIVE MODELS FOR ALZHEIMER’S DISEASE. Innovation in Aging2(suppl_1), 102-102. doi: 10.1093/geroni/igy023.382

Nie, L., Zhang, L., Meng, L., Song, X., Chang, X., & Li, X. (2017). Modeling Disease Progression via Multisource Multitask Learners: A Case Study with Alzheimer’s Disease. IEEE Transactions on Neural Networks and Learning Systems28(7), 1508-1519. doi: 10.1109/tnnls.2016.2520964

Stallard, E., Kinosian, B., & Stern, Y. (2017). Personalized predictive modeling for patients with Alzheimer’s disease using an extension of Sullivan’s life table model. Alzheimer's Research & Therapy9(1). doi: 10.1186/s13195-017-0302-6

Tang, E., Harrison, S., Errington, L., Gordon, M., Visser, P., & Novak, G. et al. (2015). Current Developments in Dementia Risk Prediction Modelling: An Updated Systematic Review. PLOS ONE10(9), e0136181. doi: 10.1371/journal.pone.0136181

Wang, T., Qiu, R., & Yu, M. (2018). Predictive Modeling of the Progression of Alzheimer’s Disease with Recurrent Neural Networks. Scientific Reports8(1). doi: 10.1038/s41598-018-27337-w