Wireless healthcare is booming in areas such as sport monitoring, elderly care, remote patient monitoring, and animal lifestock. Through wireless connected wearables and environmental monitoring wireless radar, activities and health signs can be monitored. For the analytics of this time series of raw sensor and radar information, machine learning (ML) methods have grown tremendously, but there are concerns on their transparancy. ML models are often viewed as black boxes, making it difficult to understand how they work or contribute to their forecasts, resulting in instability, lower robustness, lower fairness, biased results, and challenging generalizability. We propose to achieve improved transparency and fairness using Explainable Artificial Intelligence (XAI) methods, which can help wireless health and activity developers and designers create AI systems that meet specific needs. We plan to use XAI to gain the trust of industry players in Flanders, who are currently hesitant to use ML in their products or solutions. Furthermore, we aim to investigate how cutting edge XAI techniques, wireless activity recognition, affordable wearables, and connected healthcare can address practical challenges, including understanding decisions on heterogeneous time-series datasets and considering the privacy concerns of the explanations. Finally, we will use these explanations to design new active learning models with XAI in the loop to enhance the model's performance.