Tremendous strides have been made in recent years in the development of Alzheimer’s disease (AD) biomarkers for use in evaluating the progression of AD. However, the dynamic nature of AD progression limits the ability of a single biomarker to effectively and accurately quantify the risk of disease progression. In this study, we integrated well-studied functional, structural, molecular (cerebrospinal fluid [CSF]), and cognitive biomarkers, obtained from the Alzheimer’s Disease Neuroimaging Initiative 2 (ADNI 2) datasets, and used the event-based probabilistic (EBP) model to estimate their optimal temporal ordering sequence (TOS) (S^optimal) and stage the risk of AD progression.^1-3Subject and Image Acquisition. Using data from the ADNI 2 database, we selected 144 subjects, all of which have 10 AD biomarkers: three region-based resting-state functional magnetic resonance imaging (R-fMRI) functional connectivity indices (FCI) from the hippocampus (HIP^FCI), the posterior cingulate cortex (PCC^FCI), and the fusiform gyrus (FUS^FCI); two gray matter concentration indices (GMI) from the hippocampus (HIP^GMI) and fusiform gyrus (FUS^GMI); two CSF biomarkers of Aβ1-42 and p-tau levels; and three cognitive markers of MMSE, ADAS-Cog, and AVLT scores. Given that a set of N events, E1, E2, …, EN, is measured by N biomarkers’ value (x1, x2, …, xN, respectively), the TOS of events, S = {s(1), s(2), …, s(N)}, is calculated by a permutation of the integers 1, …, N, with the formula $$$p(X│S)=∏_{j=1}^J∑_{k=0}^Np(k)\left[∏_{i=1}^kp(x_{ij}│E_i ) ∏_{i=k+1}^Np(x_{ij}│¬E_i )\right]$$$. $$$p(x_{ij}│E_i ) $$$ and $$$p(x_{ij}│¬E_i ) $$$ are the likelihood of measurement given that event E_i has and has not occurred. k is the stage number in sequence S. To search the optimal sequence, S^optimal, among the many possible sequences, we employed a greedy algorithm to improve processing efficiency. CARE Index. The numerical order of biomarkers in the S^optimal can be used to measure disease progression from one stage to the next. Therefore, we defined the number associated with each biomarker event as a “score” and the collective scores as an “index.” We refer to this as the index for characterizing Alzheimer’s disease risk events (CARE index) and distinguish it from clinically defined AD stages (e.g., EMCI, LMCI, and AD). Optimal Ordering of Events (Fig. 1A). The first two disease events are represented by two functional biomarkers: increased HIP^FCI (1) and decreased PCC^FCI (2). The next two are CSF biomarkers: decreased Aβ_1-42 (3) and increased p-tau (4). The subsequent events are a mix of cognitive biomarkers (decreased MMSE [5], increased ADAS-Cog [6], and decreased AVLT [7] scores) as well as the gray matter concentration biomarkers (decreased HIP^GMI [8] and FUS^GMI [9]). The last event is increased functional biomarker FUS^FCI (10). The bootstrap results (Figure 1B) show event uncertainty with three clusters (1 & 2, 3-8, 9 & 10). Association of CARE Index with Clinical Stages (Fig. 2). CARE index can be calculated for individual subjects (Fig. 2A). For all subjects, we obtained a distribution of CARE index scores for subjects regardless of each subject’s clinical stage and plotted them in Fig. 2B (CN & AD) and Fig. 2C (EMCI & LMCI). Statistically, Fig. 2D shows the median CARE index scores of the CN, EMCI, LMCI, and AD groups were 2, 4, 6, and 9, respectively. The CN group exhibited a lower CARE index score than the EMCI, LMCI, and AD groups. The AD group showed a higher CARE index score than the EMCI and LMCI groups. In addition, the EMCI group showed a lower CARE index score than the LMCI group. Robustness. Through repeated measurements (Fig. 3), the CARE index’s consistency reached 89% with a slope of 1.04, which is slightly more than 1, indicating a significant intra-subject repeatability with a trend of slow disease progression. The major finding of this study is that when multiple AD biomarkers are temporally ordered, functional abnormality in the HIP and PCC networks is the earliest event in the preclinical phase of AD, even antedating detectable CSF Aβ and p-tau abnormalities. This finding sheds light on the link between preclinical AD status and its symptomatic onset and can be applied to accurately identify progressive AD trajectories, given the condition that disease onset is insidious and no single biomarker serves as a predictor for future cognitive decline.