Using conditional inference forests to examine predictive ability for future falls and syncope in older adults: Results from The Irish Longitudinal Study on Ageing

A scoping review of machine learning models to predict risk of falls in elders, without using sensor data

OBJECTIVES

This scoping review assesses machine learning (ML) tools that predicted falls, relying on information in health records without using any sensor data. The aim was to assess the available evidence on innovative techniques to improve fall prevention management.

METHODS

Studies were included if they focused on predicting fall risk with machine learning in elderly populations and were written in English. There were 13 different extracted variables, including population characteristics (community dwelling, inpatients, age range, main pathology, ethnicity/race). Furthermore, the number of variables used in the final models, as well as their type, was extracted.

RESULTS

A total of 6331 studies were retrieved, and 19 articles met criteria for data extraction. Metric performances reported by authors were commonly high in terms of accuracy (e.g., greater than 0.70). The most represented features included cardiovascular status and mobility assessments. Common gaps identified included a lack of transparent reporting and insufficient fairness assessments.

CONCLUSIONS

This review provides evidence that falls can be predicted using ML without using sensors if the amount of data and its quality is adequate. However, further studies are needed to validate these models in diverse groups and populations.

The syncope core management process in the emergency department: a consensus statement of the EUSEM syncope group

The European Society of Cardiology issued updated syncope guidelines in 2018 which included recommendations for managing syncope in the emergency department (ED) setting. However, these guidelines lack detailed process-oriented instructions regarding the fact that ED syncope patients initially present with a transient loss of consciousness (TLOC), which can have a broad spectrum of causes. This study aims to establish a European consensus on the general process of the workup and care for patients with suspected syncope and provides rules for sufficient and systematic management of the broad group of syncope (initially presenting as TLOC) patients in the ED. A variety of European diagnostic and therapeutic standards for syncope patients were reviewed and summarized in three rounds of a modified Delphi process by the European Society for Emergency Medicine syncope group. Based on a consensus statement, a detailed process pathway is created. The primary outcome of this work is the presentation of a universal process pathway for the structured management of syncope patients in European EDs. The here presented extended event process chain (eEPC) summarizes and homogenizes the process management of European ED syncope patients. Additionally, an exemplary translation of the eEPC into a practice-based flowchart algorithm, which can be used as an example for practical use in the ED, is provided in this work. Syncope patients, initially presenting with TLOC, are common and pose challenges in the ED. Despite variations in process management across Europe, the development of a universally applicable syncope eEPC in the ED was successfully achieved. Key features of the consensus and eEPC include ruling out life-threatening causes, distinguishing syncope from nonsyncopal TLOCs, employing syncope risk stratification categories and based on this, making informed decisions regarding admission or discharge.

Effect of 10-Week Whole-Body Vibration Training on Falls and Physical Performance in Older Adults: A Blinded, Randomized, Controlled Clinical Trial with 1-Year Follow-Up

Whole-body vibration training (WBV) training has shown positive effects on bone strength, muscle strength, and balance, but the evidence on fall prevention is not yet persuasive. This study aimed to evaluate the effectiveness of WBV training in preventing falls and improving physical performance among older adults at fall risk. The study was an assessor- and participant-blinded, randomized, and controlled 10-week training trial with a 10-month follow-up. One hundred and thirty older adults (mean age 78.5 years, 75% women) were randomly allocated into the WBV group (n = 68) and the low-intensity wellness group (n = 62). Falls were prospectively collected using monthly returned and verified diaries. Physical performance was evaluated at baseline before randomization, after the intervention, and follow-up with established methods. The data were analyzed on an intention-to-treat basis. Negative binomial regression was used to estimate the incidence rate ratios for falls, and Cox regression models were used to calculate the hazard ratios for fallers. Between-group differences in physical performance were estimated by generalized linear mixed models. The retention rate was 93%, and the mean adherence to the WBV training was 88% and 86% to the wellness training. Sixty-eight participants fell at least once, and there were 156 falls in total. In the WBV group, the incidence rate of falls was 1.5 (95% confidence interval 0.9 to 2.5) compared to the wellness group (p = 0.11). The hazard ratio for fallers in the WBV group was 1.29 (0.78 to 2.15) (p = 0.32). There was no between-group difference in physical performance after the training period, but by the end of the follow-up, WBV-related benefits appeared. The chair-rising capacity was maintained in the WBV group, while the benefit disappeared in the wellness group (p = 0.004). Also, the 0.5-point difference in short physical performance battery (SPPB) score favored WBV training (p = 0.009). In conclusion, progressive side-alternating WBV training was feasible and well-tolerated among fall-prone older adults. During the one-year follow-up, WBV training was associated with improved physical performance but did not prevent falls compared to chair-based group exercises.

Older adults with slow sit to stand times show reduced temporal precision of audio-visual integration

Sustained integration of sensory inputs over increased temporal delays is associated with reduced cognitive and physical functioning in older adults and adverse outcomes such as falls. Here, we explored the relationship between multisensory integration and a clinically relevant measure of balance/postural control; Sit-to-Stand Time, the efficiency with which an older adult can transition between a seated and a standing posture. We investigated whether temporal multisensory integration was associated with performance on the Five-Times Sit-to-Stand Test (FTSST) in a large sample of 2556 older adults (mean age = 63.62 years, SD = 7.50; 55% female) drawn from The Irish Longitudinal Study on Ageing (TILDA). K-means clustering was applied to FTSST data, yielding three clusters characterised by fast (mean = 10.88 s; n = 1122), medium (mean = 14.34 s; n = 1133) and slow (mean = 18.97 s; n = 301) sit-to-stand times. At wave 3 of TILDA, older adults participated in the Sound Induced Flash Illusion (SIFI), a measure of the precision of temporal audio-visual integration, which included three audio-visual stimulus onset asynchronies (SOAs): 70, 150 and 230 ms. Older adults with the slowest sit-to-stand times were more susceptible to the SIFI at the longest SOA (230 ms) compared to the shortest SOA (70 ms) relative to those with the fastest times (p = 0.02). Older adults who take longer to repeatedly transition from a seated to a standing posture exhibit an expanded temporal binding window for audio-visual events, supporting a link between multisensory perception and balance/postural control in ageing.