**6. Discussion**

The target of this work was to present a low-cost device that can be attached to commercial canes for long-term load monitoring, so it can be used for condition assessment. The proposed device has been built and tested with its target users. Results have proven that indeed: (i) it provides continuous load monitoring; (ii) it does not affect cane ergonomics; and (iii) battery allows long-term use without recharging. Additionally, results show that the load on the cane is correlated with gait speed, which is a clinically reported condition estimator.

Tests were carried out in two senior facilities in Cordoba, Andalusia. Selected volunteers required a cane for mobility. Thus, we could check load estimation when the system operates under dynamic loads through different pathological gait cycles. We also obtained users' gait speed to check that estimated load is related to speed, i.e., condition. We did not notice any difference in users' gait when they were using the modified cane, nor they reported any change. Hence, we validated that changes in ergonomics were not significant. Also, we did not need to recharge the cane at all for the whole test duration, so we also validated that it can be continuously used for long-term monitoring. Load results were coherent with the hypothesis: people with more significant disabilities consistently bear more weight on their cane and walk slower. The device provides dynamic loads during the gait cycle. We checked that load variation was larger for people in poorer condition and that they increased their load on the cane when they started getting tired.

The proposed system is not as reliable as other load estimation devices, like treadmills or pressure plates. However, those systems are constrained to specific installations, so they do not provide information while users perform their ADLs. Our module has three main advantages when compared to similar cane-based systems. First, it is a low-cost add-on module that can be easily attached to any commercial cane, rather than a stand-alone system that may present significantly higher costs and lower availability. The module has been designed to avoid changes in ergonomics and/or center of gravity of the original cane. Second, we have designed a novel sensor module that combines two low-cost load sensors at different depths into a single piece to increase their range. Optimal depth differences have been heuristically estimated to provide the best results even under extreme loads (up to 45 kg fully supported on the cane). Third, the system has been extensively tested under different loads to grant at least 3 days of operation without battery recharge, including power required by the communication module. Information can be gathered using any BT-equipped portable device like a smartphone or a tablet. We rely on an induction battery recharge system to facilitate its use to people with visual impairment. The measured error under static load is under 0.14 kg, which is suitable for the average load that our target population bears on their canes.

We expect that a significant number of these modules can be produced at a reduced cost to monitor specific target groups for long-term clinical gait analysis in hospital facilities and/or care centers. Unlike other equipment, the proposed system may be carried by users anywhere, to outside and inside environments, for days, so it could provide meaningful information about behavior trends and condition changes in people with specific disability profiles.
