Date of publication XXXX 00, 0000, date of current version XXXX 00, 0000

 
𝑹𝒆𝒄𝒂𝒍𝒍 =
𝑻𝑷
𝐓𝐏 + 𝐅𝐏
(𝟐)
𝑭𝟏 = 𝟐 ×
𝑷𝒓𝒆𝒄𝒊𝒔𝒊𝒐𝒏 ∗ 𝑹𝒆𝒄𝒂𝒍𝒍
𝐏𝐫𝐞𝐜𝐢𝐬𝐨𝐧 + 𝐑𝐞𝐜𝐚𝐥𝐥
(𝟑)
 
The following table 2 illustrates our Machine learning 
performance on the MIMIC IV dataset on clinic event 
classification using Flower Federated Learning techniques. 
We investigated the performance of various machine learning 
models, including Random Forest, Logistic Regression, 
Support Vector Machines (SVM), AdaBoost, and Gaussian 
Naïve Bayes, in a federated learning setting. The models were 
tested with different numbers of clients (3, 5, and 10) and 
communication rounds (5, 10, and 15). Our goal was to assess 
the impact of these factors on the overall performance and 
determine the most effective combination for clinical event 
classification. 
Table 2. Evaluating the Performance of Machine 
Learning Models in Federated Learning with Varying 
Rounds and Clients. 
The results of our study indicate that the highest accuracy 
across all machine learning models was achieved when using 
10 communication rounds and 5 clients in the federated 
learning environment. This optimal combination was 
observed for Random Forest, Logistic Regression, Support 
Vector Machines (SVM), AdaBoost, and Gaussian Naïve 
Bayes in the context of clinical event classification.
ML model 
Number of 
rounds 
Number of 
clients 
Train 
acc 
Test 
acc 
Random Forest 
5 
3 
97.7 
94.3 
10 
5 
98.9 
98.9 
15 
10 
97.2 
90.3 
Logistic 
Regression 
5 
3 
93.2 
90.1 
10 
5 
94 
92.3 
15 
10 
92.3 
89.3 
SGV 
5 
3 
70.1 
65.3 
10 
5 
75.4 
60.3 
15 
10 
70.3 
68.3 
AdaBoots 
5 
3 
97.6 
90.3 
10 
5 
97.4 
92.3 
15 
10 
90.1 
87.2 
Gaussian 
5 
3 
80.1 
80.3 
10 
5 
89.7 
78.3 
15 
10 
82.3 
74.3 

---

Ruzaliev R: 
Federated Learning for Clinical Event Classification Using Vital 
Signs Data 
2 
VOLUME XX, 2023