86. Eye-gaze Guided Multi-modal Alignment for Medical Representation Learning