def partition_data(dataset, num_clients):
    """
    Partitions a dataset into `num_clients` subsets.
    This is just a placeholder. Implement a more sophisticated partitioning strategy
    (e.g., based on medical specialty, patient demographics) for a real application.
    """
    data_per_client = len(dataset) // num_clients
    remaining_data = len(dataset) % num_clients
    partitioned_data = []
    start_index = 0
    for i in range(num_clients):
        end_index = start_index + data_per_client + (1 if i < remaining_data else 0)
        partitioned_data.append(dataset[start_index:end_index])
        start_index = end_index
    return partitioned_data