AutoImmune-BiAttnNet: A Novel Hybrid Attention-Enhanced Bi-LSTM Framework for Autoimmune Disease Prediction

The immune system inadvertently targets
healthy tissues in autoimmune illnesses, which are
complicated and long-lasting problems. It can be difficult
to make an early and differential diagnosis of these
conditions because they frequently exhibit overlapping
clinical symptoms, such as joint pain, exhaustion, and
systemic inflammation. To minimise permanent tissue
damage and improve treatment outcomes, these illnesses
must be accurately classified and predicted in a timely
manner. Despite the enormous potential of longitudinal
numerical data from patient records, the majority of
previous research has been on either imaging data or
static snapshots of clinical markers. There is an urgent
need for intelligent models that can capture the relational
and temporal dynamics of autoimmune disease
development, given the wealth of temporal clinical data,
which is frequently gathered over several visits. We
suggest AutoImmune-BiAttnNet, a novel deep learning
model that combines a Bidirectional Long Short-Term
Memory (Bi-LSTM) network with a Hybrid Attention
Mechanism to accurately predict multiple sclerosis (MS),
lupus erythematosus (SLE), and rheumatoid arthritis (RA)
using numerical clinical data in order to address these
issues. The model can learn disease-specific progression
patterns thanks to the Bi-LSTM component, which
records
bidirectional
consecutive clinical visits.
temporal
connections
from